13.2 Beta Drivers Leaked :)

[LtMatt]

Cheers Matt

No problem.

 

[wolvers]

Wolvers, that looks far from fine to me (not having a dig). Spikes of over 150ms and even over 200ms in one place shows it isn't.

If I have misunderstood your post and that isn't your chart, my apologies.

You're right to question those m8, and they are felt in game. It's because vsync was off, they are momentary spikes in frame time. This goes when vsync is on and isn't the constant variations in frame time that TR reported. What TR was reporting was something that may give the user a constant stuttering in game, and is something I've measured before in other games and posted here.

Here's an example from DX:HR, the only one I seem to of kept the results from.

That's the kind of thing that people often say they get in multi-gpu, except worse because it felt nasty in game. If I get the chance to test Skyrim again with vsync on I'll post the results.

Fair enough. I need to do some frame times with these 310.90 drivers to see if I get the same. They feel horrid to me in BF3 and Far Cry 3, so possibly Nvidia are prone to this frame latency issue.

I will run a game I consider smooth and compare to a game I don't and compare.

That's the best way to prove what you think you feel as stuttering in the game. If you want to make your own graphs, you can import the .csv files that fraps creates into Excel. Those one's I've done were imported into Google Docs.

All you need to do is a simple equation that calculates the difference between the first frametime and the next, then drag that down through the whole column. Highlight the results and plot it in a line graph.

Edit; I've changed the scaling on my Skyrim graph to match the one on TR.

 

[wolvers]

See above Wolvers, there are rumours going around that TR's 7950 was faulty.

They've repeated the results with the new card though it would seem.

The results on the following pages come from the MSI card. As you'll see, its performance under the Catalyst 12.11 beta drivers is very similar to what we saw from the Sapphire, with the same latency profile and the same intermittent spikes caused by high-latency frames.

Their graphs back this up.

 

[Gregster]

Done a quick test on 3 games (5 mins each). This is for comparison and sorry if anybody think I am jumping in a thread and derailing it.

Test games were Far Cry 3, BF3 and Dishonoured. Ran the test for all 3 on 310.90 and then reinstalled 306.23 and just ran BF3 (felt very smooth). Only Dishonoured felt smooth on 310.90 and I think this shows. I gamed at 5760*1080, as this is how I game always and didn't want anything to alter. No overclocks on anything eith, everything including the CPU is at stock.

Far Cry 3 results:

Dishonoured results:

BF3 results:

And finaly, BF3 on 306.23 results:

Make from this as you will but from this, Far Cry 3 is horrendous and you can see frames up and down all over the shop. BF3 not so bad but in no way as smooth as 306.23 drivers. When I ran BF3 on the 306's, it was very smooth and I ran the operation swordbreak bench we did (for 5 mins of it anyway).

In no way am I trying to argue or do the mine is better or worse, just putting this up for comparison.

 

[wolvers]

Have you tried with one GPU?

 

[Gregster]

Have you tried with one GPU?

No mate. I did this purely out of my own curiosity because drivers have always been sweet till the 310.xx. No stutter at all in any game but 310's felt stuttery, so hence why I wanted tot est the latency issue. Nobody reporting issues with one GPU, only multi cards are having problems.

 

[LtMatt]

Greg how did you get the other things to work with frafs bench viewer? When i drag the fps or minmaxavg results there it says invalid column or something.

 

[Gregster]

Greg how did you get the other things to work with frafs bench viewer? When i drag the fps or minmaxavg results there it says invalid column or something.

It just did Matt. I have time, fps and rank results. I only had frametime ticked in fraps, maybe this is it?

 

[wolvers]

When you drop a frametime.csv into it, it works out the fps from that. There's a tab at the top for switching between the graphs.

Just been testing out PS2 @1440p.

I've got the flip queue size set to 1 in RadeonPro and it's as smooth in game as it looks in the graph. I don't remember feeling those spikes in all honesty, not like in Skyrim.

 

[LtMatt]

5 minutes from sleeping dogs @ 1440p with vsync. No fps limiter. I was driving around the various cities at high speed. 7970 @ 1125/1563.

This felt super smooth, so i was surprised to see it getting up to almost 50ms as i didn't feel any lag spikes at all.

I'm tempted to try a 'Wolvers' and mess about with the flip cue size. I'm under the impression AMD default is 3?

 

[wolvers]

Yeah, default is 3 IIRC butI wouldn't expect that to stutter Matt. The vertical scaling on my PS2 graph above (can't believe no one noticed it was TESV again ) is a little bigger so it looks smoother. That's one of the things you have to watch for in those TR reviews. TBH, I don't think they fully understand how to interpret this data properly yet.

Here's what it looks like if I make the scaling the same as yours.

It's only 60 seconds so again the difference in the horizontal scaling changes how it looks. I'll do them for 5 minutes from now on.

 

[LtMatt]

You can set the benchmark in fraps to end automatically after 5 minutes, its very handy.

 

[wolvers]

Yeah, the default is 60 seconds and I just left it at that.

 

[LtMatt]

Yeah, the default is 60 seconds and I just left it at that.

Set it at 300, or longer if you're feeling ambitious.

 

[humbug]

Skyrim. Phenom II x6 1090T - Gigabyte 7870.

Catalyst 13.1

Frame latency.

Frame Rate.

Rank.

The highest frame time is 42ms, the lowest is about 10ms, average is around 18ms (60 FPS = 16.6ms)

I don't think that a frame time of 42ms is something that you can actually see a difference in.

 

[tommybhoy]

Thought I would add this here:

Installed 13.2 last night.

13.2 beta never lasted long though as I woke up this morning to see the wee man playing Spore-it was artifact city although he was playing it just the same, so back to 13.1's, but thracks did say not to use them as they were not complete.

Fair play to AMD with all the betas, but not much use when you end up rolling back due to the various small niggles you get, but they are betas at the end of the day.

FC3 is smoother on the 13.2's(but still the checkerboard bug in 3D that appeared from the betas) but the cut scenes are juddering again, Hitman has a massive improvement too-~60fps average fully maxed@1080p-at the expense of the high fps spikes and Skyrim has been given the smooth treatment too.

 

[Something Else]

so are the 13.1 drivers worth the install?

 

[humbug]

so are the 13.1 drivers worth the install?

There just the same as 12.11, and pretty solid for me, not put a foot wrong and i have just finished a 5 hour BF3 marathon.

 

[Boomstick777]

so are the 13.1 drivers worth the install?

Yes ! Beta 13.2 drivers even more so .

 

[LtMatt]

Thought I would add this here:

Installed 13.2 last night.

Thanks for the update tommy. I'm sticking to 13.1 WHQL as they seem very stable until the 13.3 beta's are released.