20-Way Radeon Comparison With Open-Source Graphics For Steam On Linux Gaming

[sthalik]

Why is it relevant in the case of information lackage, then?

For the purpose of PTS environment, measured process can be launched via sudo as a normal user, if that's the issue. Or the other way around, the supervising PTS process can be launched as root via sudo.

In any case providing data's good. Not just graphs. That tests are reproducible and verifiable, yes. Still, there's lot of effort involved in methodology of setting up the test environment. Since you have the data, please provide it. That is, samples, not graphs.

[ModplanMan]

I did not ignored your said, just suggested how you might improve performance . BTW. we don't know:

1) Did you run this particular benchmark?
2) Is it on the same resolution?

And if you have gddr5 6450 model, you are safe to add overall +30% performance on top of ddr3 6450 model results... If that is the case, then that basically have nothing to do with a driver, but with nature of hardware

Fair enough, I apologise, but your suggestion was not really related to what I was saying. I already knew that HyperZ could gain extra performance, I was trying to point out a disparity between my cards performance and the benchmark.

Don't have the card to hand, but the box displays DDR3 and not GDDR5 as it's memory. Specifically it's this one:

http://www.sapphiretech.com/presentation/product/?cid=1&gid=3&sgid=1086&pid=1166&lid=1

And a 30% increase would still not account for a 15 fps drop (read: double performance of what's being measured here).

Edit: my native resolution is 1440 x 900, but I've tested it on a 1080p screen and it's about the same - the resolution difference between the 2 would also not account for the performance difference. I can't say I've run this benchmark, but again I've played the game enough to know it hits 30 fps consistently.

[curaga]

Edit: my native resolution is 1440 x 900, but I've tested it on a 1080p screen and it's about the same - the resolution difference between the 2 would also not account for the performance difference.

It would. 2MP vs 1.3MP is a significant difference, 1.6x more pixels to push over that slow VRAM.

[Martin76]

I have a laptop that has an ATI 5470 DDR3 and Intel i3-370m.
The problem is bad reclocking.

aticonfig --odgc --odgt
Default Adapter - AMD Mobility Radeon HD 5000 Series
Core (MHz) Memory (MHz)
Current Clocks : 157 200
Current Peak : 750 800


Team Fortress 2 with fglrx 14.2, 1920x1080

13 fps

Max GPU core and memory frequency informed: 300 Mhz

Reboot. Test again.

32 fps

Max GPU core and memory frequency informed: 750 and 800 Mhz

in both cases aticonfig says the same.
I have to try with open source driver.

[dungeon]

Fair enough, I apologise, but your suggestion was not really related to what I was saying. I already knew that HyperZ could gain extra performance, I was trying to point out a disparity between my cards performance and the benchmark.

Don't have the card to hand, but the box displays DDR3 and not GDDR5 as it's memory. Specifically it's this one:

http://www.sapphiretech.com/presentation/product/?cid=1&gid=3&sgid=1086&pid=1166&lid=1

And a 30% increase would still not account for a 15 fps drop (read: double performance of what's being measured here).

Edit: my native resolution is 1440 x 900, but I've tested it on a 1080p screen and it's about the same - the resolution difference between the 2 would also not account for the performance difference. I can't say I've run this benchmark, but again I've played the game enough to know it hits 30 fps consistently.

Ah OK, i redid and recheck earlier articles with radeon and fglrx for example:

http://www.phoronix.com/scan.php?page=article&item=amd_am1_discrete&num=1

I have that Radeon 8400 on Athlon 5350 APU, according to fglrx performance are expected to be quite similar between that and 6450 .

Now with current opensource driver stack even performance si evolved greatly in recent months for Athlon 5350, i run Xonotic Ultra @ 1920x1080 and compared with this recent results for 6450:

http://www.phoronix.com/scan.php?page=article&item=amd_open_bestwatt&num=4

There 6450 has 11.55 fps and i have 25fps on 8400 (with hyperz it is 30fps) , so yeah that really can't be proper for 6450 and because of that i am now sure you are right .

Someone with the 6450 card need to compare those results with Michael's, maybe his setup is borked somehow with that card (but 5450 now also looks like it might in same problem), might be also be something specific with CPU maybe those pstate he uses, etc.. in the end if problem does not appear on other people systems it might be driver regression which needs to be bisected .

[dungeon]

Stupid 5 min limit . But yeah point is ModplanMan, it looks like halfed perf

[dungeon]

Stupid 5 min limit . But yeah point is ModplanMan, it looks like halfed perf

Might be all cards results are halfed , yeah there is posibility of that too ... i don't have all the cards