Latest Intel graphics drivers (15.22.50.64.2209) resulting in a drop in the WEI score

[code65536]

After updating to the latest Intel Sandy Bridge graphics driver that was mentioned in the Anandtech news (v15.22.50.64.2209), I saw a drop in the WEI score on both of the computers that I installed it on.

Specifically, if you look at the raw benchmarks that Windows ran (if you run the assessment from the command line it will show all the gory details of the assessment), the two "Direct3D Geometry" scores (one for D3D9, one for D3D10) just plummeted. One from 110F/s down to just 40F/s, and the other from 90F/s down to just 30F/s.

Downgrading to the April driver fixed the problem. Updating again to the September driver killed the benchmark again.

I know that WEI is just a synthetic benchmark and that this does not necessarily translate into real-world performance, but it is still troubling to see, given that this driver supposedly improves performance.

One computer has a Core i5-2400 (so it's the "low-end" 6EU Intel graphics); it's a coding/development machine, but I do some light gaming on it every once in a while. The other is a server with a Pentium G620T (so I don't really care about the graphics there, but the results on that machine are identical to what I saw on the i5).

Anyone else seeing this problem?

Edit: To run the WEI assessment from the command line, open an elevated command prompt and run:

winsat formal -restart clean

At the very end of the entire suite of tests, it will give a summary of the raw benchmark numbers. The scores that fell off a cliff are the two "Direct3D Geometry Performance" scores.

 

[mnewsham]

the WEI scores can change just by running internet explorer... dont pay attention to anything it say, does it feel any different? Does it look any different? Would ANYONE using it notice? If not WHO CARES?

 

[code65536]

It's not just a matter of the point score. But one of the underlying tests that goes into calculating that magical point score is showing dramatically different results. Consistently. On multiple systems. In multiple tests. This isn't some minor burp resulting from a background app.

 

[Lonbjerg]

WEI score?
Really.
You couldn't find an actual real world app or game to test out instead?

WEI is just as useless a 3Dmark...

 

[code65536]

I don't have any strongly GPU-guzzling apps to test with (it's why I'm using the integrated GPU ). And again, I'm not that concerned about the numeric score. But when one of the synthetic benchmarks that Windows runs to arrive at its GPU score drops by two-thirds... well, that's a significant anomaly, wouldn't you say? I mean, we're not talking about a minor blip that's within the margins of error or even something like a 10 or 20% drop... this is a huge drop...

 

[mnewsham]

I don't have any strongly GPU-guzzling apps to test with (it's why I'm using the integrated GPU ). And again, I'm not that concerned about the numeric score. But when one of the synthetic benchmarks that Windows runs to arrive at its GPU score drops by two-thirds... well, that's a significant anomaly, wouldn't you say? I mean, we're not talking about a minor blip that's within the margins of error or even something like a 10 or 20% drop... this is a huge drop...

what you arent understanding is that while the driver was updated WEI probably wasnt, it is probably only noticeable in WEI, so in everything else it wont matter.