• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Arrow Lake Builder's thread

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

It's ironic and almost LOL funny at the same time that Intel is blaming console optimizations (Zen 2) for lackluster performance in games and almost hinting that even on Windows, game developers are optimizing games for Ryzen 😀

I wish they would write a juicy technical blogpost about how these optimizations tank performance on their CPUs and what the Intel Binary Optimization Tool is doing to win back that lost performance.
 
igor_kavinski said:
I wish they would write a juicy technical blogpost about how these optimizations tank performance
Click to expand...
It's really funny they're saying this now. They're lifting games from an older baseline to something like x86_64 v4. But most optimizations they're doing would work well if not better on Zen 4/5 as well...
With NVL they can do more that Zen 6 probably won't have.
 
Obviously we need to wait for all of the independent reviews to see how Intel's claims hold up. But so far it is damned promising for salvaging 1851 as a gaming platform. As long as the benchmark claims are not scuffed e.g. using faster ram with the refresh vs OG and Raptor, 15% average uplift in a 38 game suite should make the 250K faster than the 14700K. For $300 that is not too shabby. Not all of the gains are from iBOT APO either. The ones that have the biggest gains however seem to be using iBOT, and are older games few play anymore. Shadow of the Tomb Raider, Hitman 3, Borderlands 3, Far Cry 6. Which invites speculation as to how much the list will grow with Nova not far off. Or will the same optimizations for ARL=S be for future generations?

Also cool they warranty up to 8000MT/s with the boost bios profile to eek out a little more performance yet in memory sensitive games like the Spider-Man series. Unfortunately, DDR5 pricing is going to be the turd in the punch bowl.
 
DAPUNISHER said:
Obviously we need to wait for all of the independent reviews to see how Intel's claims hold up.
Click to expand...
No, on technical merits alone any iBOT game should be excluded from comparison metrics until the binaries are extracted - though it sounds like it might be dynarec. Then that should be tested to see if it can work on Zen 5 and - hilariously - Raptor Lake for a fair comparison. Again, I expect RPL to be ahead of ARL again if given the same code to chew through.

Recompiling can introduce new bugs not present in the original (or sometimes unintentionally fix bugs present in the original). It's a path to some funny shenanigans.
 
Last edited:
gdansk said:
though it sounds like it might be dynarec.
Click to expand...
That's going to need a few gigabytes of RAM allocated for the purpose of caching recompiled instructions. I wonder if that cache will grow as needed or get saved to disk for each game and reloaded on next launch. But if Intel is doing the recompilation statically at their end and downloading the game cache with APO on demand, that's probably a better solution, though disk space intensive and lacks flexibility for recompiling stuff we want to test ourselves. I really wish they would explain the conceptual underpinnings of how their solution works without obviously giving away their secrets.

gdansk said:
Then that should be tested to see if it can work on Zen 5 and - hilariously - Raptor Lake for a fair comparison. Again, I expect RPL to be ahead of ARL again if given the same code to chew through.
Click to expand...
I think Intel will go out of their way to prevent that. They have already taken down a lot of opensource stuff that was benefiting Zen architecture more than their own CPUs.
 
gdansk said:
No, on technical merits alone any iBOT game should be excluded from comparison metrics until the binaries are extracted - though it sounds like it might be dynarec. Then that should be tested to see if it can work on Zen 5 and - hilariously - Raptor Lake for a fair comparison. Again, I expect RPL to be ahead of ARL again if given the same code to chew through.

Recompiling can introduce new bugs not present in the original (or sometimes unintentionally fix bugs present in the original). It's a path to some funny shenanigans.
Click to expand...
Interesting take. My initial reaction is to disagree. *insert why not both gif here*

I think it is perfectly fine to do the standard bigger bar better comparison, then have a section highlighting performance with the APO suite for potential buyers edification. Vendors restricting features to certain hardware is nothing new. If Intel wants to use APO to help boost arrow and not provide it to older hardware or competitors, it seems like business as usual? If not, where am I going wrong in my thinking on this?
 
DAPUNISHER said:
Interesting take. My initial reaction is to disagree. *insert why not both gif here*
Click to expand...
It's akin to letting Intel get away with bogus SPEC-specific optimizations in their compiler. But gamers are a peculiarly self-sabotaging bunch (RIP Radeon) so why not let it descend into AMD and Intel both "working with" (paying) different game developers to distribute optimized versions of their games to win in some benchmarks.

I guess in a way Microsoft is already doing dynamic code optimization when emulating x64 (still exclusive for Qualcomm at this point) so why not.
 
Last edited:
gdansk said:
It's akin to letting Intel get away with bogus SPEC-specific optimizations in their compiler. But gamers are a peculiarly self-sabotaging bunch (RIP Radeon) so why not let it descend into AMD and Intel both "working with" (paying) different game developers to distribute optimized versions of their games to win in some benchmarks.
Click to expand...
Slippery slope fallacy.

It is just marketing strategy to make chart bar bigger. However, APO has failed to move the needle since its inception. No reason to think it is going to suddenly gain traction. And few are going to build arrow with Nova probably a year away and not socket compatible.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top