ATI 5870 gets a 6.0 in gaming and graphics in Windows?

Right now, my windows experience assessment says that I got a 6.0 for graphics and gaming out of 7.9. Is this normal?

Id say thats normal, 7.9 is the highest rating you can get. Also windows experience is not a real way to measure performance, many will tell you to get a real benchmark or record some FPS in games to get a measure of performance.

The 5870 should definately show up as 7.9, its waaaaay faster than a 260 or other single gpu card. Im guessing its a driver problem.

Although Windows Experience Index is not the best tool to measure performance, my single HD4850 with 512 MB of ram scores 7.3 @ 675 MHz core and 993 MHz memory.

Your 5870 should score way more! Try refreshing maybe or update.

5850 should be 7.9 lol something about how the hell this score is done should be looked into very conflicting scores.

I've got 6.8 on a stock clock 9800GT.
Yours is wayyy lower than what it should show.

Update the scores or update the drivers.

I don't understand all of you claiming it "should be higher." How is it even computed, if you all know that it "should' be higher then please tell me what you base this on, other than how your card got score x? We can compare benchmarks with each other.. but it is rediculous to claim random BS marks mean anything at all (I'm sorry, did I mean MS marks?).

Run a game, if it performs poorly then we have something to talk about. If your FPS is fine then ignore the pointless microsoft numbers. They are just a very VERY rough outline. They are also tiered, with random hard caps such as "If card has feature x limit score to y."

Though I'd like to now exactly what the score is computed from.

My 4890 used to show up as 7.5 since last week when i updated my ATI drivers the windows experience index has been showing it as a 6.0, i believe thats the problem for whatever reason the drivers made windows think its worse.

That is unscientific garbage.

You are telling them the sky is falling because an arbitrary number doesn't fit with less than a dozen other folks.

The MS numbers are a scale to give you an idea; they are NOT a performance representation in any regard.

This is not a 3dmark program where the flops and FPS are converted to a score. There are hundreds of things at play here. The scores are capped based on dx version, based on detected specs, and any number of things I have no hard numbers on.

Unless you can tell me "this is the equation they use" you cannot go around saying it is good or bad, as there is no way to be sure. We know what fps is, we know roughly how a 3dmark score is calculated, and we don't know the same for this.

If the games get good performance there is no problem, if one wants to be sure then run a bench mark. There are far too many things that can change the index score for anyone to care about it. Perhaps a driver release will come along to make it higher, but as it doesn’t actually represent anything what does it matter if it is one or nine? It is simply a guide line, how much is based on performance vs feature compatibility?

The stdev is huge on the performance index, which is normal for something like this. The OP's deviation is well within normal for the performance index. To give him/her the impression that it is a huge problem is ridiculous without more information. It may mistakenly cap a score one day, an update may change how it is calculated the next. This is no basis to formulate a conclusion. This is not like a benchmark where we know version numbers, exactly how the scoring is done, etc. Unless that is known one cannot compare his score to anyone else’s.

The op asked if it was a problem, the answer is that it probably isn't but he should run some benchmarks of games he cares about to be sure. Which is what I said to do in my first post.

If one is concerned about performance, use a performance benchmark and we can use that to say if there is a problem or not. Using the index is akin to using what might be evenly spaced lines your buddy wrote on a paper to measure how long something is (without units) while a Vernier is sitting on the table.

As MS and driver updates come along all the time that (I’d guess artificially) raise and lower the score it is simply not something to worry about. Provided it is not so low that aero features won’t turn on there is no problem. If it is that bothersome to you you are free to look for updates to change it, but there is no point making it into a big deal and giving someoen who may not understand teh situation teh idea it is an indication sometihng is seriously wrong with his rig. I’d suspect it is an issue with Microsoft. Besides, it would not be hard to hack the system and give yourself whatever arbitrary number makes you feel safest at night.

I just wish some logical reservation was used for things like this.

Would it surprise you that much of the index scoring was being done without clocking the cards up to 3d clocks? Given that this happened all of the time in beta I’d guess that it happens still, from time to time. I suppose ATI could release a driver to force 3d clocks whenever index is open. I could never prove if it was just detecting the current clocks and saying “hey, these are lower than they should be, cap the score at x” or actually running tests while in desktop mode. At any rate, you can always remove the power play features from the bios to get 1337 windows scores..

Some measurements are more useful than others. When one form has an inherant uncertainty of more than +/-20% why would we use it for anything but what it is meant for, a very rough idea. I can get a pretty good idea how much gas is left in my tank based on how logn I figure I've been driving, but it doesnt mean I don't use the fuel gage and my trip meter.

The OP has a legitimate concern, his score is on the low end of the normal deviations from this system. That is step one, "I think there is a problem." But what he is being told as step two is that there IS a problem. When he should be running some benchmarks and posting system information. The logical thing to do when you get a hunch that is based on a very inaccurate, non precise measurement is to measure again with something you are confident in.

Yes I realise what you were saying. I should have known better than to quote your post directly, I'm sorry for that. I know you are not exactly claiming the sky is falling... my bad for laying on too much hyperbolae. I merely think you, and many others, are taking WEI far too seriously.

However, people use this system for more than it is supposed to be used for. It is a nice indication that something might (I stress might) be wrong. But there is no need to tell one that something "is" wrong until one actually does the correct tests. Many of the posts are stating that his computer must be broken based only on the WEI score, which is ridiculous given that no one here has any idea how it is calculated, or if that calculation remains constant on different versions of the OS.

There are just too many variables in WEI for it to matter. I simply get sick of this kind of measurement being used to rationalize problems; it should only be something that might make you run a second test. It is in no way something that any conclusion can be drawn.

As for the "normal" WEI for a 5870, as I don't know if it takes into account other parts of the system, etc. I can't say what normal is. But the most common complaint on Newegg about the 5000 is that "My WEI went down." Certainly this could be a driver problem... or any hundred other things. I don't know. But I would be comfortable in saying that in almost all of these situations there is nothing wrong, and the score is just being tallied wrong (if not wrong.. then strangely). I certainly don't have an ample sample size, but newegg has more than this post, and it seems there is far more variation. Besides that, there was onyl one person in this post that responded with thier own 5870 score.. how can you derive any deviation from a sample of 2? Every opther psot was a collectino of shoulds based on unsimilar rigs. Which is the basis of my "unscientific" comment. A sample of two with no rig info to determine if CPU has any bearing on WEI doesn't realyl help all that much.

By the way, Kool-Aid references are usually to imply that one is brainwashed by the team. It probably makes more sense to say that I need to take off the foil hat as it were. I am aware I went overboard a tad... But it really is an issue when conclusions are drawn from the wrong data. I don't want the OP thinking that just because his score is lower there is something wrong with his card. They need to understand that it doesn’t mean anything/much and they need to run other tests to have any idea if something is actually wrong.

Anyway, I appologize for getting carried away. I hate WEI (and many other 'colloquial' measurements) for the same reasons I currently hate records being kept in anything but ISO date and 24h time.. I can understand how both may seem rediculous to many. Irritations come and go...

The first time I ran that performance rating, I also got something around 6.0, and was monitorying ATI Overdrive which showed the core clock at 157MHz and Memory at 300MHz the whole time.

I ran it again with ATI tools? (I forget name as it's no longer installed) and set the clocks to 850 and 1200 respectively and it scored 7.9.

The test doesn't force the card out of idle timings, and as a result really messes with your score.

Had the same problem with my 5870. Decided to reload my os (win 7 ultimate 64-bit) and the score went from 6.0 to 7.8. didn't reload it because of this issue. Was tired of not having enough space on my ssd(80G).

Well my brand new Dell XPS 9000 had the same issues. So I updated the driver the newest off the ATI website...and WEI stayed at 6. So I called Dell and said...what the heck is going on! They pointed me to a brand new BIOS update that just came out for the XPS 9000. Updated the bios...reinstalled the drivers. And Shazam! My WEI index for all graphics categories is now 7.8.

Hope that helps.....

Alll of this text for a windows 7 index score?

Really?

This topic has been closed by OvrClkr