Here's Why Radeon Graphics Are Faster On Linux 3.12

[schmidtbag]

If you set the governor to performance and send the benchmark results to me, I am interested. If you set the governor to the minimum CPU frequency, I am interested. If you set the governor to ondemand, I am not interested, because it would be a waste of time for me to try and reproduce your random results which might not be reproducible on my machine at all. In that case, you should probably be sending your results to the cpufreq team and not me.

Perfectly fair request IMO. But, the underlying issue is the CPU should not have downclocked this much in the first place. I'm not saying this is the radeon driver's fault - I'm pretty certain this was an issue with cpufreq this entire time. In Windows or Mac, you don't have to mess with the governor (or whatever they call it) and their tests, to my knowledge, are not skewed. I understand why you don't want test results of ondemand, but the matter of the issue is that shouldn't be skewing the results in the first place. In other words, while disabling ondemand proves more reliable results, the tests should come out overall the same with ondemand enabled. I'm thinking kernel 3.12 is the first time the ondemand governor is working the way it SHOULD have.

On the other hand, as ChrisXY has mentioned, many of these governors are obsoleted anyway.

[oibaf]

Ubuntu 13.10

It would be nice if Ubuntu can backport that commit in 13.10 (3.11 kernel).

[ChrisXY]

So the benchmarks that showed the improvements were with intel cpus: http://www.phoronix.com/scan.php?pag...12_major&num=2

How come they even use ondemand? Has cpufreq not been superseded by pstate for several kernel versions now?

[marek]

You are not interested as a driver developer ? So if a user complains about performance problems, do you just tell him "I don't see it so there is no problem" or do you maybe try to understand what is going on ? You should be glad that someone is testing an average user configuration to spot problems that would be blamed on you.

If you set the governor to performance and send the benchmark results to me, I am interested. If you set the governor to the minimum CPU frequency, I am interested. If you set the governor to ondemand, I am not interested, because it would be a waste of time for me to try and reproduce your random results which might not be reproducible on my machine at all. In that case, you should probably be sending your results to the cpufreq team and not me.

[matyas]

Indeed, the OS should magically read average Joe's mind to determine what exactly he wants...

/facepalm

...and apparently most of the settings in software are completely useless and should be eliminated. The default way or now way. Ever. No, not even in that one special case that would be important to _you_.
Similar to the world of Gnome 3^TM
/sarcasm

[prodigy_]

Average joe configuration should be used IMO. Also the user MUST NOT have to care about stuff like that. It should just work.

Indeed, the OS should magically read average Joe's mind to determine what exactly he wants...

/facepalm

[Adarion]

To add my worthless opinion:

Well, things do happen as they do. I never really wanted to speculate where it came from but had hoped that it would be the sum of changes in the free driver stack.
Anyway.

Michael probably used a really fast PC to limit bottlenecks from that side and to actually benchmark the GPU. But then it is of course wrong to use any non-performance governor (performance = which keeps the CPU untouched and at full speed).
Still, for a real-world benchmark you just got to use ondemand since 99% of people will exactly use that one. Everything else would not make sense for all normal people. So at least that would be the effect most people would see. And it is good that a close look was done to find out what exactly caused it.

So Michael you might learn from that and check with a full-speed-CPU governor again if something looks too strange. Or keep in mind any other possible larger influences on general performance. Still I think testing also on normal-world machines and normal configurations is a fair thing since most people do not have an i7+SSD+16 GB RAM or something combination.
And please make these text boxes with the testing config larger.

[s_j_newbury]

Luke, while technically a graphically very simple GPU accelerated game might be well be non-CPU limited at the extreme, I'm not sure what you're trying to measure. It could well be you're hitting the GPU fillrate limit, exceeding the internal bandwidth of the card and at such high frame-rates I can't see how it's meaningful.

[Luke]

AMD Phenom II now fine with "ondemand" governor even if Bulldozer is not

On my FX8120, both Linux 3.11 and Linux 3.12 suffer a GPU performance hit when running the "ondemand" governor in Critter, the only non-CPU limited game I have.

Earlier I posted my FX-8120 results showing that the difference between "performance" and "ondemand" governors has narrowed but is still significant. This evening I tested the results of the two governors on a Phenom II X4, a 955BE with a Radeon HD5570. Again the high-framerate, non CPU limited game "critter" was used, it can heat that little card all the way to 70C-and run at about 730fps peak on either governor setting.

On that Phenom II, I found no discernable difference between leaving the governor in "ondemand" or putting it in "performance," no difference at all. Sometime I will have to see if that also applys to video edit rendering.

[helland]

On demand apparently does not correspond to its name

I seriously cannot believe the amount of finger-pointing going on here, and sadly its making me lose a lot of respect for people who I otherwise thought were humble and hard-workers.


It is NOT by any means Michael's fault why the results came out this way. He is not obligated to run under a different governor, just as he isn't obligated to use some obscure distro, make some minor kernel tweak or tweak the drivers for individual tests - the point of these benchmarks is to show what the average person will/should encounter from everyday upgrades. If the governor is known to be problematic then fine, switch out of it, but if the CPU is actually underclocking due to a lack of stress, that really gets me to think the governor is not the problem.

What Michael is obligated to do is benchmark using the most typical/average software setup, and a hardware setup that has the lowest probability of skewing the results (meaning, his choice of CPU, mobo, RAM, and storage were fine for testing GPUs, because all of those parts are good enough that they SHOULDN'T be a bottleneck). If you want benchmarks for the utmost highest possible results, you're in the wrong place and always have been. Even if this website was strictly benchmarks and nothing else, no single person would ever have the time to set up some of the silly or unrealistic requests here.

I'm not (yet) blaming the driver developers either, since they were being affected by an outside source.

HOWEVER

The CPU Michael used was better than almost anything AMD has to offer. That being said, it is absolutely unacceptable for the drivers to be THAT held back by the CPU, even in its low-freq state. This could mean that APUs are behind in performance simply because the CPU isn't fast enough. I'm not bashing AMD CPUs either - AMD CPUs are fully capable of playing most modern games without being maxed out.

But like I said, I'm not yet blaming the driver developers. I think tests should be done with catalyst the HD6870 (because that had the greatest performance impact) and see how much of a difference that makes. If, while testing catalyst, the 3.11 ondemand vs 3.12 ondemand has a performance impact less than 5%, that's where I think the open source radeon drivers are the blame.

I fully agree with this. I think the way Michael has done the benchmarks is the correct way to do it, and he has done an amazing job in unraveling this mystery.

As far as I can see it's the cpufreq-subsystem that is the fault here. Just read the name out loud a couple of times: ON DEMAND. That would indicate that when there is a demand for cpu-power, the governor should make sure that the demand is met. That's sorta indicated in the name of the governor, isn't it? Apparently that has not been the case. As far as I can see, a governor named ondemand should deliver max frequency if that is demanded to perform the tasks, and therefore there should be no major difference between the performance governor and the ondemand governor.

And marek's comment on not being interested in the result of a benchmark without the governor set to performance, I firmly disagree in. As far as I know marek works for AMD? Therefore he should be interested in his graphics driver performing well under "vanilla" conditions, although that does not stress his work to the max. He is, after all, contributing to a product that has to compete to be the better option, vanilla settings or not. Just because one is not to blame for a problem, does not mean one is not affected by it, and therefore should not care.

"Oh, look. One of the machines in my gym has a wire that's almost about to break. Whatever, that's not my problem. They should probably fix that, but I won't inform them". Wire breaks, I'm not able to use the pec dec until they get spare parts. Not my fault, but still I'm affected.