Intel Lists 36 Variants Of Haswell Graphics

[phoronix]

Phoronix: Intel Lists 36 Variants Of Haswell Graphics

Thanks to patches surfacing on public mailing lists on Monday that continue work on the open-source enablement of Intel Haswell graphics under Linux, we have a look at the different variants of Haswell. Somewhat surprising is that there's 36 different IDs representing the next-generation Haswell products with different flavors of integrated graphics...

http://www.phoronix.com/vr.php?view=MTE1NTQ

[Figueiredo]

Michael, it would be really great if you could do some periodic articles with benchmarks of mesa/gallium drivers across intel/nvidia/amd. I'm very curious about, and I assume other people too, which hardware puts out the best performance with the open source drivers.

As Intel closes the performance gap of its integrated graphics and also improves their mesa drivers, they may be getting pretty competitive for those looking only at open drivers. This information might be more relevant to people buying new hardware than the performance difference between open/closed drivers.

The Steam early adopters will certainly appreciate.

Just my 2 cents...

Cheers

[Michael]

Michael, it would be really great if you could do some periodic articles with benchmarks of mesa/gallium drivers across intel/nvidia/amd. I'm very curious about, and I assume other people too, which hardware puts out the best performance with the open source drivers.

As Intel closes the performance gap of its integrated graphics and also improves their mesa drivers, they may be getting pretty competitive for those looking only at open drivers. This information might be more relevant to people buying new hardware than the performance difference between open/closed drivers.

The Steam early adopters will certainly appreciate.

Just my 2 cents...

Cheers

I already do that, albeit in separate articles, but that's for page view reasons in order to cover my expenses...

[Figueiredo]

I already do that, albeit in separate articles, but that's for page view reasons in order to cover my expenses...

I understand. And I'm speaking only for myself here, but I think it would be great if you could compile these results in a single article afterwards. Would that reduce the ad revenue?

It would probably also spark some pretty interesting discussions, IMHO.

[curaga]

Ultrabook chips for servers? I can only think "SeaMicro", but with AMD owning them, why would they create new Intel models.

[Kano]

How about using wine+rage to test? A driver should be capable of running it. Thats where intel usually fails on linux. On win ivb is fast enough. You can run open arena or quake live on it, but don't consider that a real test of current hardware. Heaven maybe, something more demanding should be tested. Basically rage uses so huge textures that running with shared memory could be optimized even more compared to dedicated gfx card. A bit sad that there will not be a linux executeable for it, but it is still opengl, no big deal to try.

winetricks xact_jun2010 directx9

then run it. dx9 is not speed critical for this game, has nothing to do with rendering.

[log0]

I understand. And I'm speaking only for myself here, but I think it would be great if you could compile these results in a single article afterwards. Would that reduce the ad revenue?

It would probably also spark some pretty interesting discussions, IMHO.

I agree with @Figueiredo. Most of the time I am way too lazy to look for the other benchmarks to get an idea how the gpus perform relative to each other.

@Michael could you at least integrate the graphs of previous benchmark runs in new articles, let them accumulate? This would allow you to post every benchmark as a separate article but also provide an overview, add value/content to the article.

[Cyborg16]

Looks like Michael was right: ULT is the new ULV. Doesn't explain the CRW though (cryogenically refrigerated workhorse, anyone?).

[Figueiredo]

I agree with @Figueiredo. Most of the time I am way too lazy to look for the other benchmarks to get an idea how the gpus perform relative to each other.

@Michael could you at least integrate the graphs of previous benchmark runs in new articles, let them accumulate? This would allow you to post every benchmark as a separate article but also provide an overview, add value/content to the article.

It would also be pretty cool for comparing APUs/SoCs from vendors, specially if power consumption in thrown into the mix.

[devius]

I'm very curious about, and I assume other people too, which hardware puts out the best performance with the open source drivers.

IIRC Intel has the best performance with open source drivers (Ivy Bridge HD4000) when compared to low to medium power GPUs. Against the top of the line discrete GPUs it tends to lose. Of course that it's a bit hard comparing because it's not always the same cpu being used in the different articles.

+1 for APU comparison article