The Interesting Tale Of AMD's FirePro Drivers

[phoronix]

Phoronix: The Interesting Tale Of AMD's FirePro Drivers

Earlier this week we published our annual look at AMD's Catalyst driver releases from the past year. Not only did the Catalyst Linux driver this year picked up a couple new features, its driver performance had improved slightly over the past twelve months. In building up some initial test data for OpenBenchmarking.org we decided not only to do these tests on the latest consumer-grade graphics card this year, but expand it to cover the workstation performance too and to go back nearly two years in time. These results for an AMD FirePro V8700 graphics card with the monthly driver updates going back to Catalyst 9.2 are quite interesting. AMD announced twice this year optimizations to their FirePro driver software, but in reality these "optimizations" were largely unsustainable and not optimizations as much as they were attempting to address driver regressions from the past.

http://www.phoronix.com/vr.php?view=15592

[allquixotic]

Just some constructive criticism Michael, if I may?

Our Linux workstation used for carrying out this FirePro V8700 1GB driver testing boasted dual AMD Opteron 2384 "Shanghai" Quad-Core CPUs, a Tyan S2932 motherboard, 4GB of DDR2 ECC Registered memory, and a 160GB Western Digital SATA HDD. Going back to test the very old drivers that lack support for the newer versions of the Linux kernel and X.Org Server, we used Ubuntu 8.04.4 LTS as the base operating system. Ubuntu 8.04.4 LTS 64-bit was used with the Linux 2.6.24 kernel, GNOME 2.22.3, X.Org Server 1.4.0.90, GCC 4.2.4, and an EXT3 file-system. Every driver release from Catalyst 9.2 (fglrx 8.58.2 / OpenGL 2.1.8404) to Catalyst 10.12 (fglrx 8.80.5 / OpenGL 3.3.10362).

versus



Tell me that doesn't look better than a laundry list of components? And of course, comments are welcome from readers

I even picked out the Phoronix green color to give it a "fully dressed" feel

Format nitpicking aside, I love the detail in this article. This is a lot more in-depth and comprehensive (especially with all the SPECview stuff) than articles we've seen in the past. You really seem to be stepping up the ante with the focus on the business crowd, and while I think that will get you quite a following from people into GPGPU and rendering and such, please keep in mind the humble roots that made Phoronix.com popular in the first place. To sum it up in one phrase, I'd say that I want Phoronix to retain its keen interest in "3d games on GNU/Linux". Articles like this one show a very different Phoronix -- one that I believe is indeed compatible with the more fun-loving, game-focused Phoronix, provided that you (or future editors, since you said you may be stepping down) actively try to maintain that balance.

And of course I'm also interested in this article as a Catalyst user and a HD5970 owner, and I find that your results largely correspond with my qualitative experience over the course of the time I've had this GPU. Well done.

[bug77]

I would hope all these tests will finally put a misconception to rest.

You know, when a new video cards hits the market and everybody is like "well, that's the performance now, but let's wait until the drivers mature". There you have it folks: a year's worth of drivers and no tangible difference. The same goes for regular ATI cards and Nvidia as well.

[devius]

Tell me that doesn't look better than a laundry list of components? And of course, comments are welcome from readers

Very nice. Although it's not the prettiest table I've seen, it has the advantage of letting me know the type of hardware used in this article, since I usually just skip that part when I start reading:

"Our (machine) used for carrying out this (name of test) testing..."

We can only hope that the next editorial staff will solve this and other issues.

[allquixotic]

it has the advantage of letting me know the type of hardware used in this article, since I usually just skip that part when I start reading:

Exactly! This is why I wanted to tabulate it: to make it more readable. I, too, often skip over this section of the article. I don't know how many people also skip it, but it's probably a large percentage. If that many people skip it, you may as well not even include it, but then you'd get complaints from the 10% who read it every time, so the best of both worlds is to include it and make it readable.

[Michael]

Just some constructive criticism Michael, if I may?



versus



Tell me that doesn't look better than a laundry list of components? And of course, comments are welcome from readers

I even picked out the Phoronix green color to give it a "fully dressed" feel

I've posted about this a couple times in other threads to readers wanting similar tables.... It's already implemented in the Phoronix Test Suite and OpenBenchmarking.org to auto-generate nicer tables. Here's an example of a completely auto-generated one right now:



PTS takes care of figuring out everything and coming up with a table to highlight the differences. Once Phoronix.com is using the OpenBenchmarking.org-embedded graphs rather than static PNG files (within a couple weeks hopefully), those tables will be included, but I am not bothering with any stopgap measures in the meantime or anything that requires more manual work on my part.

[allquixotic]

I would hope all these tests will finally put a misconception to rest.

You know, when a new video cards hits the market and everybody is like "well, that's the performance now, but let's wait until the drivers mature". There you have it folks: a year's worth of drivers and no tangible difference. The same goes for regular ATI cards and Nvidia as well.

Not really. When you consider that newer drivers have the massive burden of not only running fast, but also correctly (because a great many applications don't render correctly or crash on old drivers, particularly fglrx), the difference is quite good. That they managed to add enough features (GL extensions, internal refactoring, etc) to enable all these new apps to run correctly, without reducing performance, is actually an accomplishment. Of course you could say, well, the driver ought to run every app out of the gate; but OpenGL driver<->app compatibility is a constant battle on all platforms, and it's really a give and take between the app developers and the driver developers: they both have to give a little to really make it work.

Aside from that, there weren't any tests where the most recent drivers got slaughtered, so the performance is consistent; but there are a couple tests where performance has gone up nicely over the course of the year. I don't think a whitewashed conclusion of "no performance improvement" makes sense: if you use applications with a performance and feature profile similar to (or identical to) any of the tests where recent releases have increased performance, then you will see improved performance. If you only run such apps, your experience will be very very different from someone else who, for example, only runs apps with a performance profile similar to the tests where performance had slightly dropped.

And no, you can't go drawing conclusions about the drivers of a completely different vendor with significantly different hardware based on an overgeneralization about the drivers of a competitor...

[devius]

I've posted about this a couple times in other threads to readers wanting similar tables.... It's already implemented in the Phoronix Test Suite and OpenBenchmarking.org to auto-generate nicer tables. Here's an example of a completely auto-generated one right now:

PTS takes care of figuring out everything and coming up with a table to highlight the differences.

Those tables need some tweaking. Too low contrast and too much information makes this particular example as hard to read as the textual description. At least make the text solid black on information that matters, like kernel version and hardware info. Automatic tables are nice, but some may need manual editing to remove some unnecessary information.

[bug77]

Not really. When you consider that newer drivers have the massive burden of not only running fast, but also correctly (because a great many applications don't render correctly or crash on old drivers, particularly fglrx), the difference is quite good. That they managed to add enough features (GL extensions, internal refactoring, etc) to enable all these new apps to run correctly, without reducing performance, is actually an accomplishment. Of course you could say, well, the driver ought to run every app out of the gate; but OpenGL driver<->app compatibility is a constant battle on all platforms, and it's really a give and take between the app developers and the driver developers: they both have to give a little to really make it work.

I was only talking about performance. Of course features will be added and bugs fixed. But the performance stays pretty much on the same level.
Instead, when a new cards hits the market and it can't match the competition on a performance level, all the fanboys go: "yeah, but the other card has been available for X months, just wait for the drivers to mature for the new card and then you'll see". It applies to both ATI/AMD and Nvidia fans.

[devius]

And on the other hand there's way too much contrast on the zebra coloring of the column and row titles. The rest of the cells have a background color that is way too dark.