Benchmarking programs?

All,
While I'm waiting for my RMA mobo to be delivered, had a question for y'all; what benchmarking programs are "must haves", for both system/cpu and video card benchmarking?
Never had reason to wonder about it with my old system, but once I finally get my new system up and running, I'd like to know exactly what kinda horses I've got under the hood, know what I mean?
Any useful guidance appreciated!

For testing of DX 9 features on pure VGA card and pure CPU test, use 3DMarks03.
For more stable testing of CPU+VGA and pure DX 8.1 and below features, use 3DMark 2001 SE and demo for stablilty test.
For full flegde DX9 advanced features, use AquaMark3.
For PC/HDD/Mem benchmarks, use PCMark 2002.
For realtime gaming test, use Quake3 or UT2003 benchmarks.

Similar Messages

Some benchmark programs not compatible w/ AMD64

Do any of the popular benchmarks not work with the AMD64 bit processor. So far, I am having trouble running aquamark (immediately says video out of sync), the coolbits registry program, and I get a Sandra Benchmark tip on the tests for the CPU that say to use their new 64 bit program, but it is not available for download yet.
At least the 3DMark2003 works.
Thanks.

An update on those benchmark problems. After finally getting my Nvidia driver straightened out (had to use driver cleaner to remove the old one), I was able to run aquamark with great sucsess. Never could get that coolbits registry to work, so I installed powerstrip to OC my video card. Powerstrip works great and has a lot more features that I will never use.
I currently have my video card OC'd at 430 core, 740 memory and is still stable in the benchmarks. I got a score of 40,919 in Aquamark and 5593 in 3DMark 2003. Its getting late and I don't feel like going any higher, but its nice to know this thing still has some potential.

Benchmark program for OSX?

G'day,
Just recently got iMac 24" for mum and dad, I would like to run benchmark tests on intel iMac, intel mini mac, this G4 933mhz, and PC (Intel dual duo E4400 2ghz running OSX86 on it) What program would it be called to do benchmark tests?
Norton used to have one called System Info which was great for non-osx. - What is it now?
Thanks
Tristan

Hi all,
for Cinebench, look here: http://www.maxon.net/pages/download/cinebench_e.html
Regards
Stefan

Benchmarks with Bios 1.20 Please Post

Hey all,
I should start by saying the new BIOS works very well for me. I was able to reach new high speeds. I used Aquamark 3D, 3DMark01 and 3DMark03 as benchmarking programs.
With the new BIOS I was finally able to get every last drop of juice out of my RAM by manually setting it at: [email protected] in the BIOS. I experimented with FSB or clock reference settings untill the crash point. Also with the new BIOS I was able to achieve higher than a 205 FSB with my manually specified timings. Previous BIOS's would crash consistently over 205 FSB with anything other than auto settings for me.
The sweet spot for me seems to be 214 FSB [email protected], it has a 100% completeion ratio in over 20 of the tests combined. I was able to complete one Aquamark 3D test at 218 FSB.
Results:
128M Aperture, FSB 214, [email protected]
3DMark01: 21,604
3DMark03: 6609
Aquamark 3D: 47,444
256M Aperture, FSB 214, [email protected]
3DMark01: 21,411
3DMark03: 6616
Aquamark3D: 47,545
128M Aperture, FSB 218 [email protected]
Aquamark3d: 48,012 CPU score of 10,010 , I only wish it was sustainable. This was the first bench I ran and it never ran that fast (FSB 218)again .
Though the 256M aperture wasn't much of an overall increase, it is still faster with DirectX9 based tests. I found this to be true with the Radeon 9700 Pro 128M and the 9800 XT 256M. I'm hoping the Nvidia guys here will test their systems for us in a similar fashion and post regarding the aperture settings and benchmark scores.
I am also curious about the way 1G of memory will handle (Muskin lvl. II [email protected]), if anyone could be so kind as to run these tests with 1G of RAM, I would be very appreciative.
Good work MSI any improvement is welcomed. Thanks again Bas for the early release of 1.20.

ok - some impressions
the memory is a litte bit faster
i can oc to 214fsb with the ram at 400mhz ddr, this will result in 428mhz ram speed @ 2-3-2-6
i can overclock to 225fsb if i limit my 400mhz ddr ram to 333mhz
this will result in 375mhz ram speed with 2-3-2-6
is there any possibility to limit the ram to 366mhz so i could run the fsb at 225 with ram limited to 366 - so the ram would run with 412mhz 2-3-2-6
would be a GREAT increase for the overclockers in general

General programming practice question (possibly re: clusters)

I apologize for this probably being an old topic or something I should have found elsewhere, but I simply didn't know what to search for.
I am tackling my first labview program of significant scale, and quickly discovering that the difficulty of wiring scales exponentially with the number of things you're trying to do. I marvel at the short and sweet pieces of code people post on here, and try my hardest to replicate the style to no avail.
If I've missed a good guide for good organizational practices when programming, please point the way!
Essentially, I'm grabbing info from a video signal via a daq card, using this to compute the location of objects in the video, and sending instructions to a robot to control its movement-- this means I have the image wires, serial wires, integer wires for pixel values, real wires for coordinates coordinates, and error wires flying around. It's not an extraordinary amount of data but it seems to be an extraordinary amount of scooting things around to make things look pretty and/or readable, every time I make a change.
Clearly use of subVIs is the way to go for each of the tasks, but I still have several pieces of data that need to be communicated from each task to the next.
Is it common practice to cluster the data even if it's unrelated, just to simplify the wiring? Is there significant overhead if you're unclustering and re-clustering it in each subVI? (I tried to write a quick benchmark program but I couldn't quite figure out how to use the profiler)
Any general veteran tips would be much appreciated!

Check out the following links: -
http://www.bloomy.com/resources/index.php#pres
Specifically here the Powerpoint presentations (five_techniques_for_better_labview_code.pps)
Some additional pointers/ resources below: -
http://forums.ni.com/ni/board/message?board.id=LVETF&message.id=1
http://zone.ni.com/devzone/cda/tut/p/id/4434
http://openg.org/tiki/tiki-index.php?page=Style%20Guide
There is a host of other information out there as well, try Google and use the term 'LabView style guide'.
Hope this helps put you on the right track.

Benchmarking or a "resources reducer" to use on arch linux

Helo pals....
Let me explain what I want.
I'm a folding at home entusiast, and, at this moment, I'm working with graysky's -ck kernel, liquorix kernel, trying to work with ice and pf ones (headers don't compiles with) and kernel rt
In a second time, the same kernels with NUMA, actually set to "off"
But, I'm looking for a resources reducer, or a benchmarking program that I could make "a load" on the system.
When I put to fold with a kernel, actually, I do it in idle, on normal using, and with "heavy processing", but I've some issues during heavy load:
I do it making some 20gb TAR files, using ark, but, when the files ends, it come back to idle... and, 20Gb uses more or less 20 minutes. 20 minutes is not enough to have a good idea about it.
I thought about compiling something during the folding, but it's not good too, cause if I receive some error during compiling, the system starts in a annoying mode (don't open something, or crash some applications) and I should restart
and, even compiling gives a "non linear" load...
I need to give the system a load about 30%, and/or 20 to 60% to do "real loading tests" could you help me?
arch linux running under 965BE, motherboard gigabyte 970A-D3, 12Gb RAM (2+2 4+4) with water cooler (antec kuhler 620A with 2 fans) and zogis 650Ti
PSU twice corsair VX550W

community/stress 1.0.4-3
A tool that stress tests your system (CPU, memory, I/O, disks)
community/cpuburn 1.4a-3
CPU testing utilities in optimized assembler for maximum loading of CPUs

What benchmark to use?

I just "wire tricked", and oc'ed my XP2400 a bit (2075mhz) :D . What benchmark program should I use to accurately get a take on my system.
Thanks

As far as actual benchmarks, as I've posted elsewhere, none of them are to be trusted, espeically when it comes to the nForce chipset.
FutureMark accuses nVidia of cheating (which is dumb because it's the benchmark writer's responsibility to follow hardware/driver trends and write accurate tests.) At the same time, SISandra is incapable of testing the nForce2 chipset under "real world" conditions, testing the superior DMA properties of the northbridge while accessing video at the same time.
For me, the best test, as far as benchmarking is concerned, is relative. Instead of looking at 3DMark's or SISandra's results as "hard and fast" figures, I just compare the DIFFERENCE of the two data sets, before OC'ing and after.
If you read Tom's Hardware, Hard OCP, etc., they benchmark under actual application/gaming situations, which is really the only fair way to measure, and darn near impossible to "cheat."
Additionally, nVidia themselves have yet to write drivers which are capable of fully unlocking the nForce2's potential, and still manage to beat VIA's chipsets and compete with Intel processors with "legacy drivers."
The true test is yet to come, and it's just around the corner, if we can keep nVidia's workforce out of the strip clubs.

Benchmarking and review sites

I have been watching THG especially of late and notice they have consistantly shown lower scores than I have on my system for benchmark scoring. As my system is not overclocked I find this to be interesting. For instance on CPU scores they show the A64-3400+ at 91xx score while mine is at 96xx. While I know this is not a huge difference I am showing scores that are higher than an FX-51 3200 or at least what THG is showing for that CPU.
While I am not complaining I wonder what everyone else's experience is on this issue. Have any of you checked you scores versus what the popular review sites like Tom's Hardware or Anandtech report? If so could you show them.
Personally I don't feel that they are giving prefference to anyone on these reviews but they may be doing something diferent that we are or maybe it is just something like memory or other peices they use in their testbench systems. Just the same it would be nice to see how we compair to them.

What CPU benchmarking program do you speak of?
One thing is, they may not be using the exact same version of testing software that you may be using.
Regarding the difference, I don't really think that the difference between your scores or their scores matters too much. Of course, I'm sure that you are using different equipment then the testing equipment they were using. It would be a different story if you put together the exact same system they used for testing, and you came up with different results.

BENCHMARK

HI, Using 3 different Video Benchmark programs I get consistently faster video "numbers" with a new iMac running the ATI Radeon HD 4670 than I do with my Mac Pro running a Nvidia GTX 285. That seems very strange. Any thoughts on this one? The GTX 285 has 4 times the video ram as the ATI card. Thanks, David

Barefeats?
It is well known that Apple's applications favor ATI, but 3D and games favor Nvidia.
Looked at threads on GTX 285 here?
Drivers aren't as mature or polished, is one assumption.

A PC buying guide for NLE (mainly Intel)

For AMD setups, disregard the mentions of sockets and the like but the general recommendations still hold.
When you need a new PC for editing purposes, you basically have three choices:
1. Buy an off the shelf system from companies like HP or Dell.
2. Buy a custom built PC from a specialist company, like ADK Pro Audio| Digital Audio workstations, Pro Audio Laptops, Pro audio interfaces, Pro audio software, DAW.
3. Build it yourself.
Solution 1 can offer good prices if you stick to their standard configurations and it is easy. But remember that these companies make their profit not on the base system, but on all the options they offer. If you follow their recommendations and upgrade memory, hard disks, or choose a better video card or a faster processor, they steal you blind.
I have nothing against HP or Dell, I have bought systems from them, but usually that was because of a great offer with all the features I needed at the time and applied to notebooks, desktops and servers.
Solution 2 is usually costly, but you get what you pay for. You get the expertise of the seller/builder in selecting the right components, he builds it, installs the software and tests it before delivery. The downside is that often these specialists have their own favorite hardware, like Matrox or Canopus and try to get that into the deal, which increases the price.
Solution 3 is usually the most cost effective and the most flexible, but the largest drawback is that you need to choose all the right components, to be aware of potential incompatibilities and build it yourself. Many see that as a major stumbling block, but really it is not difficult. Another drawback is support. You have to do it yourself.
The first steps to take are twofold:
1. What material do you want to edit and how do you want to deliver the results?
To put it bluntly, if you want to edit material, recorded with a mobile phone, a digital still camera, recorded off the internet from YouTube or similar, recorded from TV, grabbed from the screen, encoded with any unsuitable codec like DivX, XVid, or the like, or ripped from DVD, and that is your major purpose, no need to read this further, just use a consumer application like Windows Movie Maker or Magix for your editing.
This guide is for people who use a VIDEO camera and want to edit that.
So for people with video cameras, what is the main format you use? DV, HDV, DVCPROHD, AVCHD, XDCAMHD, other?
Do you deliver for the web, on BR or DVD or all of those?
2. What is your budget?
Once you have answered these two basic questions for yourself can you start the selection process.
It will require you to read quite a lot about PC's. Good general sources of information are:
http://www.tomshardware.com and http://anandtech.com
As a general rule of thumb you can say that the minimum requirements in terms of CPU, despite what Adobe mentions as minimum requirements, are:
1. For DV: Core 2 Duo
2. For AVCHD: i7 with HT or dual Xeon's 55xx
3. For other HD formats: Core 2 Quad
It generally does not pay to get the fastest CPU, unless you have an unlimited budget. Their price/performance is usually not the best and you will find a step down gives you nearly the same performance at a much lower price. And there is always overclocking, but I'll come back to that later. To give an example, the Core 2 Extreme QX6850 is currently more than 4 times as expensive as the Core 2 Quad Q9550, but does not deliver any noteworthy performance improvement. And both are quad cores.
To help you narrow down your choices, it is worthwhile to have a look at these charts:
http://www.tomshardware.com/charts/desptop-cpu-charts-q3-2008/Cinema-4D-Release-10,835.htm l
Likely a new version for Q1-2009 will come out shortly.
In deciding on your shortlist, keep in mind that for video editing FSB speed and cache size are very important. First look at FSB speed (the higher the better, so if you have the choice between 1066 or 1333 FSB, always choose the higher one if it fits your budget). Next narrow down your choices by selecting the CPU with the largest cache. 6 MB L3 cache is better than 3 MB.
Let's assume that after studying all these sites and reading a lot about CPU's, you have made a shortlist of possible candidates, that fit the minimum requirements indicated above, you have to identify what socket the CPU uses. Is it a 775 pin socket (Core2), a 1366 pin (i7 or Xeon 55xx) or a 771 Xeon 54xx or lower (old generation) socket. This will largely determine your choice of motherboard and memory type you need and as a consequence the case and cooling you need.
In general the chipset on a 775 mobo uses DDR2, dual channel memory, which is widely available and relatively cheap. On a 771 mobo the memory used is FBDIMM (fully buffered DDR2) with ECC (error correction) which is much harder to get and significantly more expensive. On a 1366 mobo the memory is triple channel DDR3 which is faster than DDR2 but also more expensive. For Xeon 55xx based boards memory used is FBDIMM's triple channel with ECC, which is hard to find and costly.
The main difference between DDR2 and DDR3 is that dual channel requires two populated memory banks for dual channel operation, whereas triple channel DDR3 requires three banks populated. That is the reason that mobo's with DDR2 usually come with 4 or 8 GB RAM and DDR3 mobo's with 6 or 12 GB RAM. DDR3 is faster (at the same specs) than DDR2. To complicate matters further, most 775 mobo's offer 4 memory slots, 1366 mobo's usually offer 6 memory slots and 1366 dual socket Xeon boards usually offer 12 memory slots.
I see you are losing interest, this is getting way too complicated. Well, I'm sorry but if you want to make an informed decision on what your next system should be, you need to understand these basics or buy a Dell or HP or even Alienware (a Dell subsidiary). And I continue to bore you with these technicalities in order to help you make a good purchasing decision.
When selecting a mobo for the CPU you have selected, read tests on TomsHardware or AnandTech to help you find the good performers that have the right set of features for your job. Good brands are ASUS, Abit, Gigabyte and MSI for single socket use, for dual socket the best one to check out is SuperMicro or possibly Tyan.
Pay attention to the features a board offers, like dual NIC's (network interface controller), dual firewire, number of SATA ports, chipset used for the SATA controller, IHCR and/or Marvell raid capability, on board sound, etcetera. Pay special attention to the board layout, where the PCI-e slots are located in relation to PCI slots. I have once built a machine where I knew I had to use a PCI firewire board and due to the location of the PCI-e slot and the size of the video card (dual slot size) the PCI slot was no longer accessible. It caused me quite some headaches to find a PCI-x firewire replacement board, because I had forgotten to have a close look at the mobo layout. Just a warning, so you don't fall into the same trap I did.
Ok, we have now decided on the CPU and the mobo. Let's have a look at memory.
General rule: Get as much as you can and don't spend it on higher clock capable versions! You will gain more from slower memory in large quantities, than from faster but less memory, even with 32 bit apps like CS4. One of the side effects of more memory is less use of a pagefile, which is far slower than RAM, so even if the application can not access more than 4 GB, your pagefile use will be much lower with more memory installed.
Memory is offered with various ratings. Often in marketing jargon these are dubbed as PC3-8500 or PC3-10600 or PC3-12800 up to 16000. This is all hype. The price tag increases significantly with higher numbers but the performance gain is negligent, in the order of 1 or 2 %. Your best bet is to get memory that is suitable for the FSB speed you have. If you have a FSB of 1333, use memory for 1333 FSB (10600), if you have a FSB of 1066, get 1066 capable memory (8500), it is a complete waste of money to get 1600 capable memory if your FSB does not support it. What is important however is to get the best timings you can find.
The lower the CAS latency, the better. Also look at tRAS (ROW-ACTIVE-TIME) results. These timing figures may give you better performance than spending money on higher clocked memory. On my system I have 6-6-6-18 timings for memory (even when overclocked by 35% and without increasing the voltage), which gives far better results than 9-9-9-27 timings. BTW, timings are usually measured in CAS, RAS-to-CAS, tRP and tRAS sequence. The lower these numbers, the better it is.
In general get at least 4 GB on a 775 or 6 GB on a 1366 mobo, but you may benefit from doubling these figures if you are multitasking, for instance having PR, AE and EN open for dynamic linking, or changing to PS for text or still manipulation.
Next: Storage. I point you to a previous guide I made which gives you the basics about disk setup:
http://forums.adobe.com/thread/427772
Next cases. I suggest to only look at big towers or server towers, depending on the mobo of your choice. For dual CPU sockets a server tower is almost always a must, decause of the use of e-ATX size mobo's and the requirement for specialized cooling, often using an air duct, which is only available in server cases. SuperMicro is a prime supplier and my personal favorite in terms of build quality, expansion capabilities and reliability. Be aware however that the power supply in those cases are pretty noisy. For single CPU solutions there are many choices, often based on looks. What is much more important however is the expansion room and the airflow in those cases. You don't want to pimp your editing machine with LED illuminated fans, transparent sides and the like. It needs to be functional. Antec, CoolerMaster, Lian Li and ThermalTake are common brands and worth having a look at. Looking at smaller cases with a sexy look will often lead to disappointment in the future, since they often lack the capability to add additional disk drives or burners internally or get so full with equipment, that cooling is going to be a real problem. Do take into consideration that you may need 1 or 2 5.25" slots for burners.
Warning: COOLING is essential for reliability and longevity of your system. We'll get to that in a minute.
If you work long form projects, have a multitude of projects or other situations that may require a large number of disks in use or for backup, it is worth to have a look at cases that offer SAS backplanes and (hot-)swappable disk cages. Often on various sites you may find entries to case modding, where you can find other disk cages that allow you to use four 3.5" disks in the space for two 5.25" slots. In my case I had 6 3.5" slots as is was delivered. I have modded my case to now contain 14 3.5" disks just on the front with room for another disk. To avoid disappointment in the future and a lot of work of rebuilding your PC into a new chassis, plan ahead for the storage requirements you may have in the next years. If you start out with 4
disks now, but expect to use 8 next year, be sure that they fit in, even if it requires different disk cages. Make sure that there is adequate cooling for the disks in the disk cages, preferably with 80 or 120 mm fans in front of the disks and use a replacable dust filter.
By this stage we have found the CPU, mobo, memory, hard disks and case. Time to have a look at video cards.
General: CS4 will not significantly benefit from a high end video card. The marketing hype wants us to believe that the nVidia
Quadro CX will give enormous benefits for encoding H.264, but that is just what it is: "Hype".
For general NLE work you are better served with more CPU power, more memory and more hard disks and possibly a better raid controller with more cache than with a video card that costs you at least $ 1 K extra in comparison to very good cards. Better spend the money where you benefit from it. SLI or CrossFire is a complete waste of money and only applies to gamers. Everybody serious about video editing will NEVER install a game on his machine so forget about those things.
If you make sure that you have at least 512 MB on a decent modern video card, you will be hard pressed to see significant differences in performance. ATI Radeon 4xxx and nVidia GT 2xx series are all well up to the task for editing. If you use GPU intensive plug-ins like Colorista or Magic Bullet you may be wise to choose for the more powerful versions of these cards, since these plug-ins are very demanding when rendering.
These video cards get very hot in practice and especially under load. If your case is very full or you have another card in the next PCI-e slot, reducing airflow to the video card, you may consider exchanging the stock cooler on the video card by a special cooler like the Accelero Twin Turbo from Arctic Cooling to keep temperatures in hand.
Now go to http://extreme.outervision.com/psucalculatorlite.jsp and use the PSU calculator to roughly decide what your power supply needs to deliver in terms of Wattage. I strongly suggest to get the PRO version (it is cheap!) but it saves you from fatal mistakes. A PSU (power supply unit) is one of the most overlooked and neglected components in a PC and very often the cause for all kinds of mysterious faults, hangs, BSOD's, restarts etcetera. Let me tell what nearly happened to me when building a new PC. I had checked the power supply requirements, checked reviews online and found one PSU that appealed to me, based on the reviews. It was a Zalman 1000 W. Great, looks good and in stock at an attractive price. I then got the pro version and rechecked, with the extra info on the various rails. The Zalman did not have enough power on the 5V rails that I needed. So I had to change the PSU, otherwise I would have been in for a lot of strange, unexplainable and foreall unexpected hangs, restarts and the like, apart from the reduced life expectancy of the PSU.
Now, we're finally getting somewhere. Only cooling is left, then all the hardware components have been chosen.
CPU cooling: Again check sites like AnandTech. They regularly perform extensive tests on CPU coolers. Be aware that test results differ widely from site to site, but some of the names to consider are Noctua, Scythe and Thermaltake. It is best to choose a vertical heatsink cooler with 'push-pull' configuration, meaning one fan in front to push the cooling air in to the heatsink and the other fan at the rear to pull out the air. That generally delivers the best cooling performance. These things can be quite heavy so they are not suggested for LAN parties. Install it and leave your PC where it is, otherwise the weight may damage your mobo. When installing the heatsink be very sparing in applying cooling paste (Arctic silver is a good
suggestion).
If you have followed my advise in getting a big tower you have room for additional fans in the case. There may not always be holes to attach screws for mounting, but what also works quite effectively is using heavy duty double sided tape to attach fans to a side, on the PSU (if it is mounted at the bottom of the case). Scythe has some very good fans that give a large airflow at reasonable RPM's and low sound levels.
Once you have built your system you need to decide what OS to use. 32 or 64 bit? The point seems moot now. 64 bit Vista is your best option currently. No driver problems anymore, somewhat decent stability and the use of all your installed memory. Windows 7 is only a RC currently, so I would not advise that on a production machine, but on another machine I would definitely try it. It is as stable as Vista is now, but drivers are still somewhat lacking. XP is out, despite the best stability on a clean and mean system, due to the 32 bit nature.
After installing the OS, first check your device manager. Make sure that all your devices are working and there are no warnings. Then update Windows, drivers and other system utilities. Then check your hardware, using CPU-Z and HWMonitor, check the cooling and voltages. Your disks should be below 35 C at all times. Then tune your Bios. Once you have completed these steps, proceed with installing your software. Start with essential system tools like Process Explorer, Beyond Compare, etc. and then continue with tuning Windows, removing sh*t you never use on an editing machine, like MSN or Games, setting unneeded services to manual, disabling Windows Defender and installing Symantec Endpoint protection if you have it, and only then start installing CS4 and related programs.
When all is well, you may consider to start overclocking. TomsHardware and AnandTech have published several articles on how to do that. If you do it right you will not lose stability and may get significant performance gains. It does require more attention to the temperatures in your case.
When I recently built a new system, my first PassMark (a common benchmark testing program) score was around 3600. After finetuning the system, optimizing Windows and overclocking, I got a PassMark score of 4733.8, which even after a month is still number two in the worldwide ranking.
For my system specs, look below, maybe you can derive some ideas from it for your next configuration:
http://www.millcon.nl/Harm/PCResults.jpg
I hope this has been informative and that you can profit from my remarks.
I'm sorry this was so long and taxed your patience to the extreme.
Further suggestions and enhancements are welcome.

I haven't the faintest idea. I only use DV, HDV and XDCAM-EX/HD and have never tried it with any (RED) 4K material. I assume it should be sufficient to handle that. But let's be honest, when I increased my raid array from 10 to 12 disks, I did not notice ANY performance gains. The limiting factor must be the PCI-e bus or the Areca controller, that just does not show any improvements from adding a couple of disks. Another thing to consider is that the various benchmark programs widely in use show quite different results.
As an example, here are two images from HD Tune Pro, one for my 2 disk raid0 on the Marvell chip and the other from my 12 disk raid30 array. First of all, HD Tune shows the wrong size, it is not 2199 GB but should be 10000 GB (at least that is what Areca claims as shown in the last picture), second it ONLY performs at around 745 MB/s transfer rate.
The most noteworthy item to look at is the rapid performance degradation of this raid0 when the disks get fuller. An average transfer rate of 165 MB/s is quite acceptable for a 2 disk setup, but seeing it go down to below 100 MB/s when the disks fill up is bothersome. It will not allow you to handle uncompressed HD at all.

Windows 7 Performance vs Windows XP

I have to come to the conclusion that Windows 7 takes up a lot of memory even with an updated processor. I have a Satellite L455D-S5976 laptop with an AMD Turion x64 2.1mHz dual core, 4gb ram, 250gb HDD. I was running Passmark which is a benchmark program for your computer. Now my laptop's benchark is at 474.4 compared to 1243 on the same processor. The only difference is that the 1243 score is based on a Windows XP x86 OS not Windows 7 x86 OS. I was looking at some specs compared to Windows 7 and it takes up a whole 1gb memory in which Windows XP takes up 64mb of memory. I have been trying to find a way to re-format my hard drive so I can run XP to no avail. I called Toshiba to activate my Windows 7 Downgrade Rights (this is for people that have Windows 7 and don't like the program) and they said that I can't downgrade unless I downgrade to Vista then downgrade to XP like a ladder. I have the Windows XP x86 install disks and key but I tried to re-format my hard drive so I can just install XP directly. When I run the command prompt I cannot seem to get the re-format going because it tells me that it needs to be dismounted so I dismount it and still doesn't let me reformat. I go to the disk management program and the format button is not even lit. I have came across the freeware HDDerase to see if I can reformat it that way. No go. So I am at my end with this. The only other option is to buy another bigger hard drive thats wiped and install Windows XP directly. I'm not sure if my hard drive is locked or if this can be bypassed so I can format my drive. Any suggestions would be appreciated. Thanks!

I installed the Intel beta video driver and ran a 1280 x 720 mkv file and plays without stuttering. That takes care of the video problem but doesn't fix the other problems that need drivers such as the camera. I also noticed that if I plug in my ethernet cable, Windows 7 doesn't automatically switch from wireless as XP does. To use ethernet, I have to disable the wireless in Device Manager. Hopefully Toshiba will be posting the rest of the software and drivers to make Windows 7 able to do all of the same things that XP is able to do.

Thinkpad's and N100's w/ 945PM chipset can't address 3G Ram but Dell Can

Since I am a programmer, I am particularly interested in having the maximum amount of Ram available under 32 bit Windows.
Unfortunately, I have to buy a T61p to enable me to use 4GB.
From a reply to a NoteBook review Forum:
http://forum.notebookreview.com/archive/index.php/t-136639.html
At work, we have both Dell XPS and IBM/Lenovo Thinkpad laptops. We choose the T60 models for their compactness/weight and reliability, while we choose the XPS M1710 and M2010 for their 17 and 20-inch screens and somewhat faster graphics chips. The ThinkPad T60 computers have problems accessing over 3GB RAM, and that is NOT because of the chipset or the OS that we use, but, it is only because of the incompetent designers/engineers at IBM or Lenovo; let me explain:
We have tried to install both 32-bit and 64-bit version of Vista, Windows Server 2003 R2 STD, Win XP and different dists of LINUX on T60, T60p, Z61, Z61p, Z61m, (X60 and T43) and Lenovo 3000 series, but they simply can NOT access memory over 3GB (X60 and T43 can not support that since their chipset seem to have accessing issues over the magic 3GB, but we tried anyway)
We also tried the same thing on Dell XPS M1710 and M2010, and they could access almost the ENTIRE 4 GB installed even with 32-bit version of Windows Vista (BTW, 32-bit Vista can access 16 GB RAM with the right chipset so...). Of course, now you can guess that it was no problem with the 64-bit version of the OS:es we tried.
The Dell XPS M1710 and M2010 use exactly the same chipset found in T60 and Z60 series (Mobile Intel 945PM Express Chipset). In December 2006, we even opened a T60p (model 2007 8JG) and compared the chipset with the Dell XPS M1710, and well, I suppose that you have already guessed that it was the same chipset in both machines.
Honestly, I hoped that we would find the Mobile Intel 945 GM or GMS chipset in the T60 laptops (those chipsets seem to have this 3GB limit), since we had a BIG argue with the support guys at Lenovo, and they came with totally silly answers why the computers can not access over 3GB RAM. They even asked me what I need 4GB RAM for!! Can you imagine that?! You can ask ANY programmer/developer who works with virtual machines (both VPC and VMWare or even Parallels) and everyone will tell you that you can not have too much RAM!!! And it IS A BIG difference between 3GB and 4GB. However...
Now, we have heard rumors that HP NX94xx also can access 4GB RAM installed, but I can not confirm it myself, since I have not seen it myself.
And now, regarding to your post about mixing 2GB SO-DIMM with 1GB SO-DIMM, I would not recommend it. The 2GB SO-DIMM memory modules are pretty cheap now, and besides, you will lose a lot of speed mixing 2GB with 1GB memory modules. Just try different benchmarking programs and you will see. If you want to see quick results without a lot of installations, try Memtest86+
However, I think that it is very rotten that IBM/Lenovo does not care about their customers and lets a lot of incompetent technicians answer the phone. I have tried to come in contact with higher bosses and engineers, but they are simply unavailable. But it must be a way that we customers can make them hear us.. I only do not know how! :-/
And so, here ends my first (and long) post in here!
From the thinkpad forum, there is another confirmation, but the author is not aware that Dell did a better job with the P chipset:
T60/T61/p memory limitations -- the definitive answer [pics]
http://forum.thinkpads.com/viewtopic.php?t=49703
update 10/25/2007: now includes T60/p information.
T60 and T60p:
the T60 and T60p are based off of the intel socket M processors. no matter how much memory you install in these thinkpads, only 3GB will be addressable regardless of operating system used. this is a physical limitation and there is absolutely no workaround. if you have a T60/p then installing 4GB total system memory is a waste. save your money, stop at 3GB, and put the savings either in the bank or toward a T61/p if you want 4GB.
quoted from official lenovo documentation:
Quote: Maximum memory capacity may require the replacement of standard component with largest supported component available. On ThinkPad systems with an Intel 945GM, 945PM, GM965, or PM965 chipset, even though it is possible to physically install 4GB of memory, the actual amount of memory addressable by an operating system will be limited to 3GB. This limitation does not exist with the Intel GM965 and PM965 with the 64-bit operating systems Windows XP Professional 64-bit and Windows Vista 64-bit Editions.
the key point to take away from this quote: the T60 and T60p are based off of the 945GM and 945PM chipsets and therefore fall under the 3GB physical limitation as stated above. even with 4GB total system memory and a capable OS, you still can't see or use more 3GB no matter what you do.
T61 and T61p:
the T61 and T61p are based off of the intel socket P processors. these systems are capable of using up to 4GB total system memory as long as you use the proper OS. below are the results from the four most popular microsoft OSes tested on my T61p.

i wrote the article over at TPF regarding T60/1/p memory limitations.
while i've read that certain dell models can supposedly access all 4GB, i have yet to see definitive proof backing this up.   from my understanding, dell wrote their BIOSes to report the maximum installed memory but the systems still cannot utilize more than 3GB regardless of what is reported.   dell is the only manufacturer with a 945-based system which claims to use all 4GB so it leads me to think that something fishy is going on.   if the 945 could use more than 3GB then you'd think lenovo, apple, HP, etc. would have been able to pull it off as well.
before buying my T61p i performed weeks of research on the subject of 3GB/4GB limitations in chipsets.   i have yet to find definitive evidence showing that a merom-based system can truly access more than 3GB because of the chipset's lack of memory hoisting capability.   santa rosa seems to be the first mobile chipset supporting memory hoisting (which is necessary for moving memory addresses above the 4GB mark so that all 4GB of physical memory can be addressed by the OS).
the simple solution is to buy a T61p.   while the T60p did have optional IPS displays, the T61/p is better in absolutely every aspect except display quality.   i've had my hands on numerous T60/p systems of friends/family/colleagues and wouldn't trade my T61p for any of them.   my next upgrade won't happen until intel makes a mobile chipset able to address 8GB+ system memory (which santa rosa very well may be able to do but we won't know for sure until 4GB SODIMMs are released for public consumption).
ThinkStation C20
ThinkPad X1C · X220 · X60T · s30 · 600

N100's and Thinkpads w/ 945PM chipset can't address 3G Ram but Dell Can

From a reply to a NoteBook review Forum:
http://forum.notebookreview.com/archive/index.php/t-136639.html
At work, we have both Dell XPS and IBM/Lenovo Thinkpad laptops. We choose the T60 models for their compactness/weight and reliability, while we choose the XPS M1710 and M2010 for their 17 and 20-inch screens and somewhat faster graphics chips. The ThinkPad T60 computers have problems accessing over 3GB RAM, and that is NOT because of the chipset or the OS that we use, but, it is only because of the incompetent designers/engineers at IBM or Lenovo; let me explain:
We have tried to install both 32-bit and 64-bit version of Vista, Windows Server 2003 R2 STD, Win XP and different dists of LINUX on T60, T60p, Z61, Z61p, Z61m, (X60 and T43) and Lenovo 3000 series, but they simply can NOT access memory over 3GB (X60 and T43 can not support that since their chipset seem to have accessing issues over the magic 3GB, but we tried anyway)
We also tried the same thing on Dell XPS M1710 and M2010, and they could access almost the ENTIRE 4 GB installed even with 32-bit version of Windows Vista (BTW, 32-bit Vista can access 16 GB RAM with the right chipset so...). Of course, now you can guess that it was no problem with the 64-bit version of the OS:es we tried.
The Dell XPS M1710 and M2010 use exactly the same chipset found in T60 and Z60 series (Mobile Intel 945PM Express Chipset). In December 2006, we even opened a T60p (model 2007 8JG) and compared the chipset with the Dell XPS M1710, and well, I suppose that you have already guessed that it was the same chipset in both machines.
Honestly, I hoped that we would find the Mobile Intel 945 GM or GMS chipset in the T60 laptops (those chipsets seem to have this 3GB limit), since we had a BIG argue with the support guys at Lenovo, and they came with totally silly answers why the computers can not access over 3GB RAM. They even asked me what I need 4GB RAM for!! Can you imagine that?! You can ask ANY programmer/developer who works with virtual machines (both VPC and VMWare or even Parallels) and everyone will tell you that you can not have too much RAM!!! And it IS A BIG difference between 3GB and 4GB. However...
Now, we have heard rumors that HP NX94xx also can access 4GB RAM installed, but I can not confirm it myself, since I have not seen it myself.
And now, regarding to your post about mixing 2GB SO-DIMM with 1GB SO-DIMM, I would not recommend it. The 2GB SO-DIMM memory modules are pretty cheap now, and besides, you will lose a lot of speed mixing 2GB with 1GB memory modules. Just try different benchmarking programs and you will see. If you want to see quick results without a lot of installations, try Memtest86+
However, I think that it is very rotten that IBM/Lenovo does not care about their customers and lets a lot of incompetent technicians answer the phone. I have tried to come in contact with higher bosses and engineers, but they are simply unavailable. But it must be a way that we customers can make them hear us.. I only do not know how! :-/
And so, here ends my first (and long) post in here!
Message Edited by charlestek on 01-05-2008 07:59 PM

Hi
Supporting greater than 4GByte isn't easy. It could be a piece of hardware in the Lenovo/IBM laptops that are preventing greater than 3GByte which might be resolved by a driver update.
Some info here: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEdrv.mspx
I agree Lenovo support hasn't been very good as they could have provided some information themselves or offered to look into it.
Regards
Phil
Note from Moderator: Converted to clickable link.
Message Edited by nonny on 01-06-2008 11:06 AM

W520 i7-2860QM with 2 memory modules and CPU-Z Reporting them as Slots 1 and 3?

Is this normal for CPU-Z (1.59, 1.60, 1.61) to report a matched kit pair of properly populated memory modules as Slots 1 and 3 in "dual" mode? One is under the keypad in the upper slot #0 while the other is in slot #1 under the chasis on the left closest to the DVD (with the black plastic frame).
Slots 2 and 4 in CPU-Z are reported empty.
I think this is okay and just the way CPU-Z is setup to report the SPD info but would like to double check out here with others using only 2 of their 4 memory slots.
I'm running a fresh clean bloatware-free Win7 Pro SP1 installed to my Samsung 840 Pro 512GB SATA-III SSD and holding at BIOS 1.36. I'm using 1600 speed 2x8GB Corsair Vengeance modules (a 2x8 kit with adjacent module serial numbers). Optimus and the integrated graphics are both disabled in BIOS and the unit never leaves its 170W Series 3 Plus docking station w/ external monitor, keyboard and mouse ..... it runs with the lid closed nice and cool, plus quiet also.
CPU-Z also reports I'm actually running at 1600 on the memory and I have thrown every torture test I know of at this configuration and have not had any issues yet over the last 6 weeks. Benchmark programs are also all on the very high end of the ranges published for W520 units ..... probably with a lot of help from that 512GB SSD SATA-III Samsung 840 Pro.
I never plan to ever use the on-board graphics, so I'm getting ready to install my other identical Corsair 2x8 kit next week and see if stability issues arise at 32GB using 4 modules ...... will hold at BIOS 1.36 unless things start happening at 32GB.
Thanks in advance for any input about the way CPU-Z reports pairs of memory modules.
Two (I used to have Five) - T61p 15.4" WS T9300 2.5Ghz units, August 08/08 Builds with FX570M Nvidia Graphics; ... One W520 i7-2860QM w/2000M Nvidia Graphics (most recent acquisition and stupidly fast); .... One - T42 4:3 15" Flexview 1.8GHz with ATI Graphics (still perfect for traveling); ... Two - T500 15.4" units both with ATI HD3650 Graphics.

Quote
Originally posted by Wonkanoby
if it drops that low rms the psu
Sorry for asking, but what's rms? I found nothing on google that seemed to fit in, but I guess you mean to get rid of the PSU and get a new (better) one.
I also ran memtest86 for 3+ hours and it seemed that one of my memorysticks was heavily corrupted.
So I removed that one and ran memtest86 again for an hour without any problems, but the volt keep dropping anyway... and the freezing-problem persists.
Hopefully I can purchase a PSU on approval and run some test.
Thanks for the quick answers guys!
edit: By the way, which brand do you recommend for the PSU? Right now I have Topower, cheapest one I could find when I bought my computer about a year ago (only bought the two extra memorysticks last week).

New W520 BIOS 1.37 has been posted:

New W520 BIOS 1.37 has been posted:
http://download.lenovo.com/ibmdl/pub...s/8buj15us.exe
http://download.lenovo.com/ibmdl/pub...s/8buj15us.txt
[Problem fixes]
- Fixed an issue where incorrect battery capacity might be shown in some battery
benchmark programs.
- Fixed an issue where the computer failed to resume normal operation from
hibernation state after running Setup Defaults by the F9 key in the ThinkPad
Setup menu.
- Fixed an issue where the Power-On NumLock function in ThinkPad Setup did not
work.
Although there is no mention of it above, BIOS 1.37 fixed the "50 second cold-boot start-up delay" issue with my Intel 510 SSD.
Hurrah!!!! and thank you Lenovo! (especially "someotherguy")
PS: However, my Kingston HyperX 1600 Mhz RAM now runs at 1333 Mhz (664.6Mhz x 2 - per HWiNFO64)
W520 4270CTO i7-2820QM Quadro2000M 1920x1080 Display 16GB RAM 2x240GB Intel 510 SSDs (RAID 0) - BIOS 1.42 - PCMark7:4,568
Samsung Series 9 15-inch NP900X4C-A03US - PCMark7: 4674
Solved!
Go to Solution.

Perhaps the memory clock limit was to address a possible cause of the SSS issue. I've been running 1.37 with Crucial 4x8 gb SODIMMs w/o any issue. They are capable of running at 761MHz but are being clocked at 664.6 Mhz. Not a big deal for me.
huberth wrote:
New W520 BIOS 1.37 has been posted:
http://download.lenovo.com/ibmdl/pub...s/8buj15us.exe
http://download.lenovo.com/ibmdl/pub...s/8buj15us.txt
[Problem fixes]
- Fixed an issue where incorrect battery capacity might be shown in some battery
benchmark programs.
- Fixed an issue where the computer failed to resume normal operation from
hibernation state after running Setup Defaults by the F9 key in the ThinkPad
Setup menu.
- Fixed an issue where the Power-On NumLock function in ThinkPad Setup did not
work.
Although there is no mention of it above, BIOS 1.37 fixed the "50 second cold-boot start-up delay" issue with my Intel 510 SSD.
Hurrah!!!! and thank you Lenovo! (especially "someotherguy")
PS: However, my Kingston HyperX 1600 Mhz RAM now runs at 1333 Mhz (664.6Mhz x 2 - per HWiNFO64)
W520, i7-2820QM, BIOS 1.42, 1920x1080 FHD, 32 GB RAM, 2000M NVIDIA GPU, Samsung 850 Pro 1TB SSD, Crucial M550 mSata 512GB, WD 2TB USB 3.0, eSata Plextor PX-LB950UE BluRay
W520, i7-2760QM, BIOS 1.42 1920x1080 FHD, 32 GB RAM, 1000M NVIDIA GPU, Crucial M500 480GB mSata SSD, Hitachi 500GB HDD, WD 2TB USB 3.0

Benchmarking programs?

Similar Messages

Maybe you are looking for