Engineering the Windows 7 “Windows Experience Index”

re: Engineering the Windows 7 “Windows Experience Index”

gkeramidas — Mon, 19 Jan 2009 12:19:57 GMT

regarding the disk score, a lot of us in the beta program are seeing low scores, i know mine was 1.9. then i and lot of others, turned of disk caching in device manger and our scores all improved. mine went from 1.9 to 5.9.

now, you also mention that it tests the primary drive. is this the drive that windows is installed on, or the actual primary on the controller? i know i had to turn of caching for all of the drives before my score improved. i tried the drive the os was on, didn't help, so i just turned caching off for all drives.

i have since turned caching back on.

re: Engineering the Windows 7 “Windows Experience Index”

Hurricane Andrew — Mon, 19 Jan 2009 16:17:31 GMT

I hardly thing that many of the Disk scores are accurate. I have an older 80GB 7200 RPM IDE hard disk with an 8MB buffer in my work PC, and in my primary home machine (on which I'm dual booting Vista/7) I have a 320 GB 7200 RPM SATA-II drive with a 16MB buffer. The IDE scores a 5.4 and the SATA a 2.9? I hardly believe that's accurate.

In addition, encoding and burning an hour-long avi to DVD-video using Nero Vision generally takes 35-40 minutes in Vista on my hardware, but in Windows 7 it was well over an hour and only about 2/3 done encoding. I could see the video was very "jerky" and hardly smooth in the preview window during the process under Win7. This could very easily be due to real disk issues.

My other test PC also scored a 2.9 with a SATA-II, 7200RPM disk, but it dual boots XP, so no comparison can be made there. I'm looking for another system with an IDE disk to test. I strongly suspect that it will "show" performance better than any of my SATA systems.

re: Engineering the Windows 7 “Windows Experience Index”

mdaria510 — Mon, 19 Jan 2009 16:40:28 GMT

If I'm being perfectly honest, I have to say I find it difficult to take the WEI seriously. Its a case where it's oversimplified to the point of being useless.

From what I can gather, there's a lot of detailed measuring that actually goes into determining each subscore. But the problem is that the user, especially use power users who are the only ones who actually care in the first place, are simply not privy to the measurements in the first place.

It's hard to take a 6.9 seriously when you have absolutely no basis for how that number was reached. As a front end, simplified method of performance metrics, its a fine idea. But there absolutely needs to be a entry somewhere where we can see the details, and it would be nice if there was a whitepaper we could look up how these scores are calculated.

What makes a good disk score? Random writes, sequential writes, free space, access time? Who knows!

What makes a good 3d graphics score? Fill rate, memory size, memory speed, number of stream processors, etc? Who knows!

There's no reason to withhold this information from those who actually care to know it.

re: Engineering the Windows 7 “Windows Experience Index”

Big Dan — Mon, 19 Jan 2009 16:52:08 GMT

I appreciate all the work that has gone into Windows 7 and as much as Vista was bashed by everyone and their mother, Vista is a good operating system.

I like the idea of system scoring. I didn't realize Vista did this until I got my new computer home, it would be nice if it were advertised more and possibly even had a sticker on the box. I doubt that will happen as sellers of low end computers are in for some tough questions.

Something that would be great is if it were scored on a basis of 10. No one scales things on a basis of 5.9 or 7.9. While I understand there are technical reasons behind those numbers, it just makes it harder for lay people to understand. Even if it was rounded off, on the basis of 10, it would be a whole lot easier for someone buying a computer off the shelf to understand.

re: Engineering the Windows 7 “Windows Experience Index”

smartpatrol — Mon, 19 Jan 2009 18:37:24 GMT

Very good points by other posters; using WEI as a reliable way to determine Software performance on a given machine is a mistake. Why doesn't my WEI go up when i shutoff memory hogging bloat features like aero?

re: Engineering the Windows 7 “Windows Experience Index”

elyoh — Mon, 19 Jan 2009 19:14:22 GMT

I strongly disagree with the notion that disk scores should/may change between Vista and 7.

In any analysis exercise such as WEI, the method of oberving and measuring should remain constant otherwise the results become meaningless! Imagine if they changed the way unemployment statistics were calculated from one period to the next! This is exactly what is happening with WEI. Any unbiased analyst would suggest the same thing.

A WEI score should be permanent unless the actual hardware is improved e.g. by firmware improvements or new components installed.

I was particularly shocked that a brand new hard drive went from 5.0 in Vista to 3.0 in Windows 7. That's a 40% reduction in score. (Is Windows 7 40% worse than Vista?) A reasonable statistical deviation would be something like plus minus 5%.

Increasing the max score from 5.9 to 7.9 should take account of the new advances in hardware so there should be no need to "downgrade/cap" the score of older hardware. The WEI max should be raised further say to 8.9 so newer hardware scores relatively better than the old without degrading the scores of the old.

If these changes are made permanent in the final release, then as the WEI will become an unreliable source of data and I thing anyone who uses WEI will be disappointed.

re: Engineering the Windows 7 “Windows Experience Index”

mrfortin — Mon, 19 Jan 2009 20:50:39 GMT

Thanks for the feedback so far. We take it very seriously. Let me see if I can address a couple common points made thus far.

On the matter of write caching, we very much believe it is a mistake that the WinEI score improves dramatically when write caching is disabled. We don't recommend disabling write caching and are working to understand how best to prevent the scores from improving so dramatically simply by disabling the cache. We do know, tha write caching typcially helps best with large sequential reading sequences and that disabling the cache prevents the build up of background work that may later interfere w/ subsequent reads. In other words, with caching disable we don't see the very long IOs that result in our capping the score at a low level.

On the matter of transparency, it is indeed our plan to disclose in great detail how the scores are calculated, what the tests attempt to measure, why, and how they map to realistic scenarios and usage patterns.

For the disk assessments, the tests are run on the disk that has Windows 7 installed on it.

On the matter of keeping the tests consistent between operating system releases, we debated this internally and decided it was best to accept some changes to address positive, and sometimes negative, issues that impact realistic scenarios. In support of this decision, I'd like to point out we had a great deal of data in our hands highlighting some common performance issues with disks, including almost all of the early solid state disks as they hit the market. Given the WinEI tests were not sophisticated enough to catch the problem, it seemed wrong for us to continue to highlight the drives as being good, or very good, when in fact they were the root of many responsiveness issues.

re: Engineering the Windows 7 “Windows Experience Index”

kfarmer — Mon, 19 Jan 2009 20:56:59 GMT

@mrfortin:

I think part of the conclusion stems from the now-established practice of bragging about one's WEI score. Instead of listing the score, you might instead describe it in qualitative terms: "Your system performance indicates that you can expect the following experience: .... To improve the experience, here are some things you might consider changing: ...."

Of course, that removes the possibility of scoring software for purchase.

re: Engineering the Windows 7 “Windows Experience Index”

confuzatron — Mon, 19 Jan 2009 21:51:13 GMT

"Why doesn't my WEI go up when i shutoff memory hogging bloat features like aero?"

For the same reason the horsepower of your car doesn't go up when you rip out the back seats.

re: Engineering the Windows 7 “Windows Experience Index”

confuzatron — Mon, 19 Jan 2009 21:53:39 GMT

"In any analysis exercise such as WEI, the method of oberving and measuring should remain constant otherwise the results become meaningless!

Imagine if they changed the way unemployment statistics were calculated from one period to the next!"

Uh... the way unemployment is counted has been updated multiple times over the years.

re: Engineering the Windows 7 “Windows Experience Index”

confuzatron — Mon, 19 Jan 2009 22:00:06 GMT

"I was particularly shocked that a brand new hard drive went from 5.0 in Vista to 3.0 in Windows 7. That's a 40% reduction in score. (Is Windows 7 40% worse than Vista?) A reasonable statistical deviation would be something like plus minus 5%."

If you read the article, you'll see that the way scores are calculated has changed to be a more accurate reflection of the performance of the hardware. Vista calculations didnt take into account pathological behaviours that MS discovered after the testing algorithms were developed.

So basically, your HDD was always a 3.0, Vista just wasnt aware of it.

re: Engineering the Windows 7 “Windows Experience Index”

elyoh — Mon, 19 Jan 2009 23:13:04 GMT

Maybe it would be sensible to upgrade the WEI in Vista via the next service pack so everything is consistent between the operating systems.

re: Engineering the Windows 7 “Windows Experience Index”

daved1948 — Mon, 19 Jan 2009 23:25:32 GMT

Why is it Windows 7 makes changes to MP3 files? Who died and left you as God or the IP Police?

I can't understand this mentality, and frankly, I don't think it's even legal, is it? How is an operating system granted rights to permanently alter files which I own and have legally made? If you truly respect Digital Rights, then you have no rights to alter files found on my network. They don't belong to you, nor may you automatically modify them.

I'd suggest you take a hard look at this idea before you get yourself into a class-action suit that'll once again, put you at gross odds with the great majority of potential users.

I don't condone this action taken by Windows 7, and I doubt many will.

re: Engineering the Windows 7 “Windows Experience Index”

elyoh — Mon, 19 Jan 2009 23:32:52 GMT

"Given the WinEI tests were not sophisticated enough to catch the problem, it seemed wrong for us to continue to highlight the drives as being good, or very good, when in fact they were the root of many responsiveness issues."

I agree that it would be better in explaining some performance issues but this will inevitablely sacrifice some consistency in comparision between systems on Vista and 7.

However put like that, it is probably more useful to be able to find the bottle necks in the system with a more accurate score. It's a shame the Vista WinEI tests for hard drives had a 40% uncertainty in their analysis on some drives.

[And regarding unemployment - it was only meant to be a loose analogy i.e. "You can prove anything with statistics"]

re: Engineering the Windows 7 “Windows Experience Index”

nwoolls — Tue, 20 Jan 2009 00:23:01 GMT

@daved1948 - "Why is it Windows 7 makes changes to MP3 files?"

Because the user has the option enabled in WMP to update media files with information from the Internet?

re: Engineering the Windows 7 “Windows Experience Index”

xiphi — Tue, 20 Jan 2009 01:58:09 GMT

@david1948, So, you don't condone it in Windows 7, but you'll condone it in Vista and XP? You do realize WMP is set to update media information by default when you use the Express Setup, right? It's not just happening on 7.

If you're so protective of your files, perhaps you should make them all "read-only".

re: Engineering the Windows 7 “Windows Experience Index”

Fault — Tue, 20 Jan 2009 02:36:41 GMT

The disk test caps are entirely counterproductive to the overall WEI score - my system can still easily handle 5.0 WEI score applications, but the arbitrary disk test caps rate my RAID 0 array at 3.0, which sets my score (as displayed in the Games Explorer) at 3.0.

I know full well to ignore the WEI, but this is going to be more than a little confusing to the intended target of the WEI system.

re: Engineering the Windows 7 “Windows Experience Index”

mdaria510 — Tue, 20 Jan 2009 05:31:42 GMT

@mfortin

I'm glad to see transparency is forthcoming, I assume this is because it is still under works.

If and when the details are revealed, I highly suggest it be done within windows through help and support or some other internal method, rather than a blog like this. That kind of information should be a click away from the actual score.

I think it would be far more interesting if there was more detailed information given rather than a general explanation of how scores are reached. For instance, if both memory capacity and speed have an effect on the score, rather than just saying that, it would be nice to see on some sort of advanced details dialog the measured read/write speeds, latency, etc.

re: Engineering the Windows 7 “Windows Experience Index”

blargsoup — Tue, 20 Jan 2009 08:29:03 GMT

At first, I wondered why would you use a scale that ends in 5.9? Just reading people complain about having their HDD score drop helps me to understand the purpose of this scale. The idea was to increase the top end according to hardware available at the time. If they made it on a scale to ten, then every update to WEI would result in a drop in score for the same hardware, and otherwise rational people would be crying.

I am particularly excited about the adjustment in HDD scores to reflect real world results (besides some oddness with write caching). People might helped to realize that the biggest upgrade you can get for a relatively modern system is a SSD of good repute.

I do not think I have come across any software that depicts minimum WEI values for proper operation besides Microsoft's general guidelines. I wonder if there are any plans to help get this introduced in the minimum/recommended requirements specifications section for new software.

re: Engineering the Windows 7 “Windows Experience Index”

Monarch — Tue, 20 Jan 2009 16:59:07 GMT

"Some drives, however, reads are often held off for very lengthy periods as the drives apparently attempt to clear their backlog of work and this results in a perceived “blocking” state or almost a “locked system”"

I can only confirm this. Currently I'm using a Vista SP1 system with two Seagate 400 GB drives and huge amounts of Outlook mail (4-5 GB). These drives get a WEI rating of 5.3 (very good).

When I work inside Outlook deleting, moving etc. mails the program regularly locks up completely and the HDD led flashes constantly. This makes the whole computer seem to be very slow.

"we synthesized many of the I/O sequences and undertook a large study on many, many disk drives, including solid state drives"

It would be great if you published this study :-)

re: Engineering the Windows 7 “Windows Experience Index”

anonymuos — Tue, 20 Jan 2009 18:29:51 GMT

Laptop:

CPU: 4.7

Memory: 4.7

Graphics: 3.2

Gaming graphics: 4.9

HDD:3.0

HDD with Write Caching turned off: 5.4

:O Please 'correct' this. Or should I turn off write caching? Will have to re-test it now using Iometer and Sandra.

Also for HDD score, do you look at the drive RPM while rating it?

I'm Very Glad The WEI Was Updated in Windows 7!

mikefarinha1 — Tue, 20 Jan 2009 18:46:36 GMT

On my Seagate 500GB (7200.10) I would get A Vista WEI score of 5.3... However I knew from daily use that my HDD was rather sluggish and I couldn't understand why it had such a high WEI score.

I was pleasantly disappointed that Windows 7 rated my HDD as a measly 2.9.

I'll now feel more confidant in purchasing a HDD based on the Windows 7 WEI score rather than the Vista WEI score.

I also want to agree with an eariler commentor that said it would be more beneficial to make the top WEI score a 10 (or 9.9). This will be more intuitive to the casual user that happens across his WEI score. It should also be required that future software depict both a Vista WEI score and a Windows 7 WEI score. That will help reinforce the idea that the scores are calculated differently.

re: Engineering the Windows 7 “Windows Experience Index”

mvadu — Tue, 20 Jan 2009 19:27:45 GMT

Toshiba satellite with MK1637GSX

(http://www.toshibastorage.com/main.aspx?Path=StorageSolutions/2.5-inchHardDiskDrives/MK1637GSXPage)

Vista WEI: 4.8

Windows 7 WEI: 2.0

No difference in this rating based on write caching.

re: Engineering the Windows 7 “Windows Experience Index”

mikedopp — Wed, 21 Jan 2009 00:53:06 GMT

Of course why are we even worrying about this? The Windows Experience Index can be altered to show any wonderful number.

Here have a look: http://weblogs.asp.net/mikedopp/archive/2008/12/25/is-your-windows-experience-index-lying-to-you-windows-experience-index-editing.aspx

This works for all versions of vista including Windows 7 and if you get really creative Windows xp as the app can be added to xp yes.

Don't forget the Buffer overflow error when using this tool. http://www.nabble.com/Windows-Vista-winsat.exe-Integer-Overflow-td16363210.html

All good times.

-mikedopp

re: Engineering the Windows 7 “Windows Experience Index”

falkao — Wed, 21 Jan 2009 00:59:54 GMT

Q9300@3.4, 4 GB RAM@1066, ATI 4870

CPU: 7.4

RAM: 5.9

Graphics: 7.9

Gaming graphics: 6.8

HDD: 5.9

When I first saw those scores I’ve thought, ah seems they already considered faster systems, good to know. Especially on Ram and Hdd with a score of 5.9 it seems that DDR3 and SSD drives are already considered. Hence the WEI makes quite sense to me.

re: Engineering the Windows 7 “Windows Experience Index”

Ryan523 — Wed, 21 Jan 2009 03:22:49 GMT

Question,

I have been testing Windows 7 on a partitoned Hard drive. of 25GB When I run the test it give me a 1.0 on my (Data Transer rate) on my Vista OS with same hard drive it gives me a 5.3 (Dual Boot)

Now the only thing I could come up with was that it was a partitoned drive... and it was only 25GB out of 120GB wich might have caused it to think the data rate is slow. But it is a question that I can not find a solution

Anybody have an idea?

Thanks Ryan

re: Engineering the Windows 7 “Windows Experience Index”

xplode — Wed, 21 Jan 2009 04:12:59 GMT

Well lets say that vista WEI is incorrect for the hard drives score, then why a lot slower and older hard drives based on IDE interface gain much higher score in Windows 7 WEI than SATAII based drives, when the performance of the IDE ones is clearly lower?

Also i can agree to lower the scores for hard disk drives in favor of SSD, but only in a condition when Windows 7 is released to manifacturers that we can buy lets say 500 GB SSD for a 150-250$ which is far from possible.

IMO there are few bugs in Windows 7 WEI scoring system which will be worked on and fixed in future builds.

re: Engineering the Windows 7 “Windows Experience Index”

pinobot — Wed, 21 Jan 2009 08:46:45 GMT

Gigabyte MA78GM-S2H mobo (ATI HD3200 DX10 WDDM 1.1) with AMD 4850e processor and 2 GB ddr 800 memory gets a Windows Experience Index score of 3,5 for aero but my Asus P5KPL-MA (Intel GMA3100 DX9) mobo with Intel 5200E processor and 2GB ddr 800 geheugengets a score of 4.1 for aero.

re: Engineering the Windows 7 “Windows Experience Index”

a_a — Wed, 21 Jan 2009 09:44:37 GMT

Hello there, the most experience I hope to be improved in Windows 7 is, please don't jump any window or dialog to the front and bring the input focus when I didn't allowed it!

This is really an evil feature, specially when I'm inputting and a window with keyboard shortcuts jumps out, one of these shortcuts will instantly occurred and I even don't have time to see what's happened! It's really causing misoperations.

And again, DON'T jump any window to the front and bring the input focus when I didn't allowed it please!

Is here is correct place to post this?

re: Engineering the Windows 7 “Windows Experience Index”

Gigaplex — Wed, 21 Jan 2009 10:07:03 GMT

"Well lets say that vista WEI is incorrect for the hard drives score, then why a lot slower and older hard drives based on IDE interface gain much higher score in Windows 7 WEI than SATAII based drives, when the performance of the IDE ones is clearly lower?"

Because those slower IDE hard drives don't suffer the same crippling performance issue, and are therefore actually faster in some situations. Besides, IDE hard drives aren't generally all that slow as the bottleneck is in the platter, not the interface.

re: Engineering the Windows 7 “Windows Experience Index”

metalhead_82 — Wed, 21 Jan 2009 10:39:33 GMT

My Scores:

CPU: 7.2

RAM: 7.2

Graphics: 7.4

Gaming Graphics: 6.0

Hard Disk: 6.0

For info, if anyone cares, the disk is a previous model Hitachi Deskstar which cost £30. :-)

re: Engineering the Windows 7 “Windows Experience Index”

Ronny49 — Wed, 21 Jan 2009 14:57:33 GMT

Computer:

MB: Asus Striker II NSE

Intel Core 2 Quad 9550 2,83GHz

ATI Radeon HD 4870X2 WDDM 1.1 driver

8GB DDR3 1333MHz

Samsung 1TB 32MB

Windows 7 64Bit

Score

Processor 7,3

Memory 7,3

Graphics 7,9

Gaming graphics 6,3

Primery Hard disk 6,0

Question: Gaming graphics 6.3 ????? Vista 5.9

re: Engineering the Windows 7 “Windows Experience Index”

hitman721 — Wed, 21 Jan 2009 22:53:57 GMT

In all the feedback on the WEI, I don't get why on certain standard Hard Drives, the write catching is turned on. On SSD's and the HDD's, where the chipset actually benefits the write catching, I could see the benefit. However, if you're hardware suffer's because of such a feature, why doesn't Windows automatically turn it off?

I get it that you want the best experience for everyone. Write Catching probably has its advantages. My point is that most folks who aren't investing in higher end HDD's and SSD's, are going to see this as a negative. If I can get a good Windows Experience without write catching, then I think this feature needs to be intuitive.

re: Engineering the Windows 7 “Windows Experience Index”

martinmine — Thu, 22 Jan 2009 00:49:56 GMT

First of all, thanks for trying the beta! It was an awesome experiece, and I wil give you ALOT feedback ;)

Annyway, my WEI if from 2.8 to 5.1(CPU 3,9, RAM 5.1, Graphics 2.8, Gaming graphics 3.0, primary HDD 2.9)

But why was my HDD score lover than Vista?

Martin

re: Engineering the Windows 7 “Windows Experience Index”

rgr — Thu, 22 Jan 2009 02:10:10 GMT

Hi Ronny49,

I'm the development manger for WinEI. Your gaming score went up because you have very well performing adapter, DX10 and a WDDM 1.1 driver. Of course, WinEI is a benchmark, but it does take other things into account than simple measured performance. Gaming graphics are a good example: to get to a 6 requires both DX10 and a WDDM driver. Without these, the gaming and graphics scores both max out a 5.9.

re: Engineering the Windows 7 “Windows Experience Index”

rgr — Thu, 22 Jan 2009 02:10:16 GMT

Hi Ronny49,

I'm the development manager for WinEI. Your gaming score went up because you have very well performing adapter, DX10 and a WDDM 1.1 driver. Of course, WinEI is a benchmark, but it does take other things into account than simple measured performance. Gaming graphics are a good example: to get to a 6 requires both DX10 and a WDDM driver. Without these, the gaming and graphics scores both max out a 5.9.

re: Engineering the Windows 7 “Windows Experience Index”

Ronny49 — Thu, 22 Jan 2009 13:18:06 GMT

Thanks to everyone who replied to my question, especially to rgr.

Although, I think I didn't aks my question quite right.

I did expect my score to be higher than 5.9 but a lot higher than 6.3 (in Windows 7)because I have one of the market's best graphics adapters.

What can I do to increase the 6.3 score?

re: Engineering the Windows 7 “Windows Experience Index”

BogdanPopa — Thu, 22 Jan 2009 19:17:23 GMT

strange...

my WEI CPU score grow from 5.0 in vista to 5.3 in 7, HDD score grow from 5.7 in vista to 5.9 in 7, memory remain the same 5.9 both vista or 7, and the aero and graphics grow much higher in 7 than vista. i guess its about your drivers conf., try with a vista driver if 7 are not avaible

re: Engineering the Windows 7 “Windows Experience Index”

JBClark — Fri, 23 Jan 2009 05:00:41 GMT

I hope the RAM score is better clamped to sizes instead of bandwidth. Bandwidth is not that important and makes comparatively little difference in real performance, however size is vital. IIRC, in Vista with just 1.5GB of RAM you could have a score of 5.9. That's ridiculous, there were games coming out that required 2GB already, and it wasn't that much. I'd expect, say, 4GB to be the minimum for a 5.0 and 8GB for an 6.0. Maybe with 2GB it should be clamped to 4.0. Now, there's the issue of needing 64-bit to address it all...

re: Engineering the Windows 7 “Windows Experience Index”

drewgrey — Fri, 23 Jan 2009 07:28:39 GMT

I think it is great that the weakest link in my computer has been identified. I look forward to seeing tests on hard drives that have a meaningful ,standardized and understandable metric. This should hold the feet of some of these slackers to the fire. I would like to see a single hard drive with 4 platters have 4 sata controllers so that I can have raid with one drive. Oh and make it a hybrid please.

re: Engineering the Windows 7 “Windows Experience Index”

mark_ms — Mon, 26 Jan 2009 08:34:07 GMT

Is there a link between calculating my WEI and the use of thumbnails by the taskbar? On 3 laptops, 1 with ATI and 2 with Intel graphics, I was getting small black squares as pop-up thumbnails on the taskbar, but after running the WEI, they started working properly.

Skewing the numbers around

xwindowsjunkie — Sun, 01 Feb 2009 19:23:20 GMT

Dump the "Windows Experience Index".

Its the most idiotic thing I've seen in Windows. It was a bad idea in Vista and its even worse in Win7. The number is meaningless. It has little to no relationship to reality.

The effort spent working on it is better spent making the rest of the OS better.

The fact that you can "game" it by turning on and off disk caching, and video performance settings and so forth invalidates it as a serious tool for any real use.

The fact that most of the people on this blog seem to think it can be affected by whether or not thumbnails and the like are turned on or off and that measurements in Vista don't match in Win7 etc means that its not a tool, its a marketing "feature".

re: Engineering the Windows 7 “Windows Experience Index”

tfm — Mon, 02 Feb 2009 02:58:04 GMT

I have just scored 2.9 on a 300Gb Hitachi Ultrastar 15k rpm SAS drive running from a 3ware 9690SA controller. In my experience this kit gives very good real-world performance. So IMHO 2.9 is a positively misleading number. I get 5.9 if I turn off write caching, which is simply perverse.

Tim

re: Engineering the Windows 7 “Windows Experience Index”

tfm — Mon, 02 Feb 2009 03:30:17 GMT

Further to my last post, if I turn off the command queuing function I can leave write caching enabled and get a score of 5.9.

But in this situation Everest reveals a lower linear and random read score (c.18MB/s) than with command queuing enabled (c.22MB/s).

So, again, I think something's awry.

Tim

re: Engineering the Windows 7 “Windows Experience Index”

tfm — Mon, 02 Feb 2009 03:34:01 GMT

PS: those low scores are for 4KB block size, obviously: the 64KB block size gives read rates of above 80MB/s.

Tim

Very odd graphics score

Gary_F — Fri, 08 May 2009 12:26:30 GMT

Windows 7 RC rated my graphics card (ATI X1700 built into the laptop) as 4.4 for “Graphics” which is pretty good. Then I was alerted to some Windows Updates including one for ATI graphics which was unexpected. After the update and a reboot I re-ran the performance test and was astonished that it downgraded my respectable 4.4 rating to a measly 2.1. I ran it again to check and it was the same! How can a Windows Update for 7 RC more than halve the score?

The “Gaming Graphics” rating remained at 3.3. So Windows 7 after the update reckons my card is more capable of 3D gaming that it is of moving pretty windows around the screen. This does not make sense either.

re: Engineering the Windows 7 “Windows Experience Index”

Hun73rdk — Fri, 29 May 2009 17:58:20 GMT

I have 3 x western digital 500GB re2 harddrives in raid0 and i cant get the disk rating over 5.9

they need to fix that faulty test

i can read with over 200MB/s and write with over 200MB/s and wlile writing and reading on the same time am on about 100MB/s R+W at the same time

re: Engineering the Windows 7 “Windows Experience Index”

saulc — Wed, 10 Jun 2009 19:44:58 GMT

I can't get the WEI tool to complete. It will start, but then it either makes no progress at all, or it dies trying to do the Direct3D 9 Aero assessment. As a result, I'm not seeing Aero. My computer has an Intel dual-core CPU and 8G of memory, and my graphics card is an ATI HD 2600 XT. How would I assess where my problem lies?

re: Engineering the Windows 7 “Windows Experience Index”

Holiday Logo — Mon, 22 Jun 2009 08:05:51 GMT

That's ridiculous, there were games coming out that required 2GB already, and it wasn't that much. I'd expect, say, 4GB to be the minimum for a 5.0 and 8GB for an 6.0. Maybe with 2GB it should be clamped to 4.0. Now, there's the issue of needing 64-bit to address it all...

re: Engineering the Windows 7 “Windows Experience Index”

Financial Logo — Mon, 22 Jun 2009 08:07:29 GMT

I would like to see a single hard drive with 4 platters have 4 sata controllers so that I can have raid with one drive. Oh and make it a hybrid please.

re: Engineering the Windows 7 “Windows Experience Index”

Engineering Logo — Mon, 22 Jun 2009 08:10:29 GMT

Gaming graphics are a good example: to get to a 6 requires both DX10 and a WDDM driver. Without these, the gaming and graphics scores both max out a 5.9.

re: Engineering the Windows 7 “Windows Experience Index”

Charity Logos — Mon, 22 Jun 2009 08:12:38 GMT

there were games coming out that required 2GB already, and it wasn't that much. I'd expect, say, 4GB to be the minimum for a 5.0 and 8GB for an 6.0. Maybe with 2GB it should be clamped to 4.0. Now, there's the issue of needing 64-bit to address it all...

re: Engineering the Windows 7 “Windows Experience Index”

Charity Logos — Mon, 22 Jun 2009 08:14:01 GMT

In support of this decision, I'd like to point out we had a great deal of data in our hands highlighting some common performance issues with disks, including almost all of the early solid state disks as they hit the market. Given the WinEI tests were not sophisticated enough to catch the problem, it seemed wrong for us to continue to highlight the drives as being good, or very good, when in fact they were the root of many responsiveness issues.

re: Engineering the Windows 7 “Windows Experience Index”

Drugs Logo — Mon, 22 Jun 2009 08:17:36 GMT

re: Engineering the Windows 7 “Windows Experience Index”

peterdrichter — Tue, 18 Aug 2009 18:50:50 GMT

First of all, Great job on Win7! Also, don't cave to the mass nitwits that want a 10.0 scale!

So, I have a question about the Disk test - I'm using RAID 10 with an intel onboard raid controller (SATA). I originally had 4 Raptor 10k rpm drives in Vista x64 - acheived a 5.9. The heat got to much, so I changed to (4) 7200 rpm seagate drives, also Raid10. Score stayed the same 5.9. Now in Win7, the score is still 5.9. I have a feeling if I went to a single drive that the score would still be a 5.9. :) On my little lenovo X301 with SSD I get a 6.2 so I know Win7x64 can do better! I'm curious about how RAID systems are taken into account. Thanks!

re: Engineering the Windows 7 “Windows Experience Index”

Hagrinas — Sat, 22 Aug 2009 22:43:01 GMT

I agree that the 10.0 makes no sense. If hardware becomes more powerful in upcoming years, Windows might decide I am a 10.0. Then there will be no way to compare my performance to anything better that comes out.

I think the current scheme already suffers from this. I am also stuck at 5.9 for RAID. It's not that I think I deserve more. It's that I had a problem with a drive and changed it to another with lesser performance. I still got a 5.9.

Capping makes it impossible to compare, since by definition, devices with more capability than others will have the same scores. 5.9 might make sense when computing the overall score, but I might have a 6.1 level instead of my previous 6.3 level of a 5.9 class. Microsoft is showing you a tree's branches but hiding the leaves. The leaves are the measurable sub-factors and hard criteria.

If MS rightly decides that devices that don't meet future spec XYZ should get a 5.9 sub-sub-score, then as long as the other sub-sub-scores (leaves) are at least 5.9 then the sub-score (branch score)will be 5.9. Anybody with a leaf less that 5.9 will get a meaningful sub-score. We want to see the tree. MS is worrying about showing the forest.

Also, the overall score will tell a user that a machine with a 7.1 for calc per sec and a 4.9 for RAM will not benefit(i.e. get a better score) by getting more/faster RAM if gaming graphics keeps the overall score at 4.5.

The caps and lack of score improvement for hardware improvements is something that should be addressed. Or a white paper could explain the details.

However I agree with the overall score being the lowest of all scores. There is still room to consider sub-scores and compare them to what a software vendor prints on the box or lists on a website.

The advantage of the low score is that you won't get ambushed by a new application that is constrained by your lowest sub-score. An average might boost things based on a high score for something that is not even a constraint for a given user.

Somebody who has a killer media center PC might never need the CPU power in the box. A gaming graphics bottleneck might not result in a perceivable difference in performance. But when that user decides that this super fast PC should be perfect for gaming, the user might be in for a shock.

Back in the mainframe days, we could set the priority for individual users as well as for applications. I met with the senior VP and told him that I would give him the lowest priority. The rationale was that it would have no effect when there were no constraints caused by contention, but he would be the first to know when they caused performance to degrade. Many administrators did the opposite, making top management oblivious to performance problems until end users were crawling in mud.

He knew that users were getting at least the performance he was, and probably more. Windows users will know that a computer will give them at least what the score dictates, and probably more.

re: Engineering the Windows 7 “Windows Experience Index”

medical tourism — Thu, 12 Nov 2009 09:58:23 GMT

re: Engineering the Windows 7 “Windows Experience Index”

markovich — Wed, 18 Nov 2009 22:06:41 GMT

On my Seagate 500GB (7200.10) I would get A Vista WEI score of 5.3... However I knew from daily use that my HDD was rather sluggish and I couldn't understand why it had such a high WEI score.

I was pleasantly disappointed that Windows 7 rated my HDD as a measly 2.9.