Bulldozer and minimum framerates...

[blckgrffn]

Have any of the reviews touched on this?

I may have missed it...

It was my impression that PII and Thuban struggled from time to time especially regarding minimum framerates in MMO's, etc.

For me, average frames @ 100+ isn't a big deal, but if Bulldozer helped out minimum frames that would still be something to tout and consider. Like the 5870 versus a 6950 in Crysis at Max, average framerates don't show how much smoother the experience is without dips into 20FPS.

Thanks! 🙂

Read the HardOCP review - interesting. Good in 2 of 3 DX11 games and punished in Civ. Hmm... Hopefully we'll see more comprehensive coverage in the near term...

 

[Vic Vega]

Well, if you decided before you review what the tone and conclusion of said review would be, you probably wouldn't include information which speaks against those items. 🙂

Speaking to PII minimum frame rates, I don't play MMOs but the games I do play, (L4D2, CoD4, BFBC2, SC2) have no problems on my system at all and do not slow down at all where it would be noticeable in game play. Game play is completely fluid. In fact, playing on my box side by side with my coworkers i7 box is virtually indistinguishable at 1920x1200. If you weren't told what the specs of the box were, you wouldn't know one was faster on paper.

 

[blckgrffn]

Well, if you decided before you review what the tone and conclusion of said review would be, you probably wouldn't include information which speaks against those items. 🙂

Speaking to PII minimum frame rates, I don't play MMOs but the games I do play, (L4D2, CoD4, BFBC2, SC2) have no problems on my system at all and do not slow down at all where it would be noticeable in game play. Game play is completely fluid. In fact, playing on my box side by side with my coworkers i7 box is virtually indistinguishable at 1920x1200.

Hah! Well, I am not buying into conspiracy theories yet, but after reading Anandtech and xbit I was surprised that all we were seeing were average framerate graphs and not even touching on minimum frames...

It could go the other way too, with the single thread performance being what it is, the dips might be even be lower than the old CPUs... I just want more data 🙂

My Q6700, which is about pound for pound with PII doesn't give me many troubles either but I usually disable vsync and put up with the tearing...

 

[soccerballtux]

when you overlcock the CPU-nB/ L3 cache the framerate dips are much alleviated

 

[Concillian]

My personal experience benchmarking games I own, minimum framerates are the ONLY thing that matters, and there are different cases that cause minimum framerate limitations. It's been a while since I did a set of extremely analytical benchmarks.

I haven't played a game that needed it in a while. Battlefield: Vietnam was the last time I did it (as I said, a while ago.) I could repeatably create two different scenarios of minimum framerates. One that hammered the GPU subsystem and only changes in the GPU subsystem could alleviate the minimum framerate issues. One that hammered the CPU subsystem and only the CPU subsystem could alleviate the minimum framerate issues. The memory subsystems would reach a saturation point pretty early, and below a threshold throughput, it would kind of just drop all framerates pretty evenly. While it was a while ago (As I recall this was with an AthlonXP and a Geforce 5900 + Radeon 9700 pro,) I think the same will still hold true, on most games today.

You could really only tell which was which by analyzing 'instantaneous' FPS and over / under clocking the three different subsystems (CPU / GPU / memory).

Site reviews give a first order approach that tells you big differences. [H]ardOCPs approach clearly gives the most information, but they all fall short of something that tells the whole story. In their overclocked vs. not overclocked, for example, they say there is no benefit, when there clearly is. There is one spike that creates the minimum frame rate. This spike repeats in all tests and doesn't budge. But several other spikes that were about the same magnitude DID budge.

It will take someone tweaking the HT speed and other junk that can be tweaked to move all the minimum spikes. Or perhaps that's a GPU related minimum FPS spike? Whatever the case it takes more than a couple days of benching and generalized comments to figure out the entire story. That kind of stuff takes time and effort that generally comes from the community rather than review sites.