Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel's Lion Cove P-Core and Gaming Workloads

chipsandcheese.com

Post reply on HN
Print view View original

91–100 of 126 posts

Re: Intel's Lion Cove P-Core and Gaming Workloads

#91

Post by jeffbee »

moffkalast wrote: post #87
jeffbee wrote: post #81

Earlier quoted context omitted.

Did you read the article?

The article literally says this very thing? > Lion Cove suffers harder with backend memory latency, but far less from frontend latency. Part of this can be explained by Zen 4’s stronger data-side memory subsystem. The AMD Ryzen 9 7950X3D I previously tested on has 96 MB of L3 cache on the first die, and has lower L3 latency than Lion Cove in Intel’s Arrow Lake platform. Beyond L3, AMD achieves better load-to-use late…

Yeah, latency. Strix has higher bandwidth but it pays for it in higher (i.e. worse) latency.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#92

Post by dist-epoch »

cogman10 wrote: post #90
StillBored wrote: post #80

Earlier quoted context omitted.

Which is just caching/locality asymmetries, knowledge of which has been at least partially integrated into schedulers for a couple decades now. It just goes to show how hard scheduling actually is. But also, you call it a 'massive' problem and its actually somewhat small in comparison to what can happen with vastly different core types in the same machine. Many of those cores also have quite large cache differences t…

I think part of the problem is that from where I stand, there's no way to tell my programming language (java) "Hey, this thing doesn't need horsepower so prefer to schedule it on little cores." Or conversely "Hey, this thing is CPU sensitive, don't put it on a LITTLE core." I don't think (but could be wrong) that C++ has a platform independent way of doing this either. I'm not even sure if such an API is exposed from…

Windows has an API for that, I don't think it's widely used though:

> On platforms with heterogeneous processors, the QoS of a thread may restrict scheduling to a subset of processors, or indicate a preference for a particular class of processor.

https://learn.microsoft.com/en-us/windows/win32/procthread/q...

Re: Intel's Lion Cove P-Core and Gaming Workloads

#93

Post by moffkalast »

jeffbee wrote: post #91
moffkalast wrote: post #87

Earlier quoted context omitted.

The article literally says this very thing? > Lion Cove suffers harder with backend memory latency, but far less from frontend latency. Part of this can be explained by Zen 4’s stronger data-side memory subsystem. The AMD Ryzen 9 7950X3D I previously tested on has 96 MB of L3 cache on the first die, and has lower L3 latency than Lion Cove in Intel’s Arrow Lake platform. Beyond L3, AMD achieves better load-to-use late…

Yeah, latency . Strix has higher bandwidth but it pays for it in higher (i.e. worse) latency.

They literally say AMD gets better latency on top of better bandwidth, are we reading the same sentences? Better = lower I would expect.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#94

Post by onli »

blacklion wrote: post #42
onli wrote: post #7

To see what that means in practice, in my multi generational meta benchmark the 285K lands currently only on rank 12, behind the top Intel processors from the last two generations (i7-13700K and 14700K plus the respective i9) and several AMD processors. https://www.pc-kombo.com/us/benchmark/games/cpu . The 3D cache just helps a lot in games, but the loss against the own predecessor must hurt even more.

Is your benchmark trustworthy? I see strange discrepancy: my "old" i7-12700K vs i7-13700K: Games: 170 vs 270 Software: 271 vs 1875 (!!!) I can believe into 170 vs 270 (though, these two CPUs are not so different!) but 7x difference in software!? Is it believable?

I found a bug in the score calculation that inflated the score, this case is closer now.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#95

Post by jeffbee »

moffkalast wrote: post #93
jeffbee wrote: post #91

Earlier quoted context omitted.

Yeah, latency . Strix has higher bandwidth but it pays for it in higher (i.e. worse) latency.

They literally say AMD gets better latency on top of better bandwidth, are we reading the same sentences? Better = lower I would expect.

They're referring to the ordinary DDR5-on-a-stick Ryzen implementation, not the LPDDR5X Ryzen AI implementation.

ETA: The only non-laptop results I have seen for the Ryzen AI place it worse than the 9950X, which has 2-channel plain old DDR5. The 4-channel LPDDR5X only wins when it can exploit the bandwidth and hide the latency, such as physics simulations. But for most applications the 9950X beats it. Hence my original point now well upthread: still waiting to see if the Strix Halo architecture is good for anything.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#96

Post by kcb »

dandellion wrote: post #50
pjmlp wrote: post #14

Earlier quoted context omitted.

I think many haven't yet grasped the future is heterogeneous computing, especially, when many desktops are actually laptops nowadays. Software working poorly in such setup means no effort was made to actually make it perform well in first place. Games requiring desktop cases looking like a rainbow aquarium with top everything will become a niche, in today's mobile computing world, and with diminishing sales and atten…

People have been making the same argument for more than a decade now, meanwhile PC gaming has only kept growing. If that was the case, why hasn't it happened yet?

Every kid and teen I know wants one of those shiny gaming PCs. Much more than past decades thanks to all of their favorite streamers predominantly playing on gaming PCs. They have half an aisle of those PCs just at Costco. Seems out of touch to suggest this is a declining trend.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#97

Post by MangoToupe »

mort96 wrote: post #8

> E-Cores are turned off in the BIOS, because setting affinity to P-Cores caused massive stuttering in Call of Duty. I understand doing this for the purpose of specifically analyzing the P-core microarchitecture in isolation. However this does make the test less interesting for potential customers. I don't think many people would disable E-cores in BIOS if they bought this CPU, so for the purpose of deciding which CP…

I wonder if there's also something to do with how the window scheduler (or allocator, or driver interaction, or some system component—I just guessed the scheduler because it seems like the most likely thing to be different enough to drive this sort of phenomenon) approaches the task. I notice remarkably smoother gameplay with the same games under wine—even, confusingly, games developed in house at Microsoft like Flight Simulator—with the same hardware.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#98

Post by CoastalCoder »

CoastalCoder wrote: post #60
pjmlp wrote: post #49

Earlier quoted context omitted.

I did mention C was supported, though if you really wanted to push them to the limit, and given the limited memory as well, Assembly was the way to go.

Fair. At the time I packed the necessary understanding to know of the GCC-emitted object code was significantly suboptimal.

s/packed/lacked/

Re: Intel's Lion Cove P-Core and Gaming Workloads

#99

Post by MobiusHorizons »

mort96 wrote: post #8

> E-Cores are turned off in the BIOS, because setting affinity to P-Cores caused massive stuttering in Call of Duty. I understand doing this for the purpose of specifically analyzing the P-core microarchitecture in isolation. However this does make the test less interesting for potential customers. I don't think many people would disable E-cores in BIOS if they bought this CPU, so for the purpose of deciding which CP…

Yep, I don't think chipsandcheese is trying to write a review of the CPU for practical purposes. Their content is more on technical microachitechture minutiae. In pretty much all of their articles they work to isolate performance characteristics of subsystems to show the differences that can be attributed to just the changes in that system gen-on-gen or between manufacturers.

Re: Intel's Lion Cove P-Core and Gaming Workloads

#100

Post by anonymars »

onli wrote: post #83
Aurornis wrote: post #78

Earlier quoted context omitted.

> I seem to remember you'd need dedicated industrial cooling for the 14700k. Those CPUs run hot, but it got exaggerated a lot online. It’s not hard to handle their heat with a good air cooler (even some of the $50 units like the Peerless Assassin) or a run of the mill closed loop water cooler. There are a lot of old notions in the gaming community that you need to keep CPUs under arbitrary temperature thresholds or t…

Hm, to keep in mind though that what the gaming community always claimed actually did happen with those processors - they disintegrated because of too much voltage (and probably heat). https://www.pcgamer.com/hardware/processors/intel-cpu-crashe... . So the "run themselves deep into the performance curves" part of these Intel processors was a disaster.

They were cooking themselves at idle too (see microcode 12F) so it's not clear heat/throttling is relevant
Post reply on HN