- Mar 3, 2017
- 1,747
- 6,598
- 136
Oh it did crash, and in doing so it died and entered the post-launch underworld. Now it rides as a phantom among us, a spectral horror.The Hype Train was supposed to crash and blow up, not keep spinning its wheels after it derailed, lol.
Certainly, but the very fact that GB also work with Windows is a prove that they pretend to be representative of PC perfs as well, otherwise they would do a specific test suite that is smartphones dedicated.Smartphone usage is often spiky.
That cant be true since there s official support for DDR 5600, what he mean eventually is that big OEMs will use only LPDDR.According to Linus Strix Point has no SODIMM support at all, which also means that Framework for instance can't use it in their laptops.
I really hope this policy will change. I don't care if this is achieved with CAMM3 or whatnot, but SKUs with expandable memory really should be supported for higher-end laptops and mini-pc's
I watched that video, and I was sure that's a mistake. I think be meant to say there aren't any laptops with SODIMM for now.According to Linus Strix Point has no SODIMM support at all, which also means that Framework for instance can't use it in their laptops.
I really hope this policy will change. I don't care if this is achieved with CAMM3 or whatnot, but SKUs with expandable memory really should be supported for higher-end laptops and mini-pc's
Only just found that out recently when it was mentioned here.Edit : That s why Geekbench do pauses between each test, that s a blatant help for Apple as their core wouldnt yield good numbers if the tests would be run without delays.
When has Zen ever had a good uncore?I'm mostly talking about STX's horrendous uncore. 1MB L3 per Z5c core, and Meteor Lake's P-to-LP-core level cross-CCX latency.
The HX 370 specs say that it supports DDR5. Does it only support soldered memory?According to Linus Strix Point has no SODIMM support at all, which also means that Framework for instance can't use it in their laptops.
I really hope this policy will change. I don't care if this is achieved with CAMM3 or whatnot, but SKUs with expandable memory really should be supported for higher-end laptops and mini-pc's
Indeed. DDR laptops are a dying breed. LPDDR has more bandwidth and is more power efficient- ideal properties for a laptop APU. A historical shortcoming of LPDDR was that it only came in soldered form, but now that's changing thanks to LPCAMM.That cant be true since there s official support for DDR 5600, what he mean eventually is that big OEMs will use only LPDDR
Zen 3 was a significant improvement in gaming perf largely due to the uncore.When has Zen ever had a good uncore?
Also that's 500% a Zen 6 target: Z5 was meant to be the core, Z6 the uncore.
They've absolutely kicked that can down the road.
Man, the cross-CCX penalty is 3 times higher than that of Renoir. It's atrocious.When has Zen ever had a good uncore?
Also that's 500% a Zen 6 target: Z5 was meant to be the core, Z6 the uncore.
They've absolutely kicked that can down the road.
There’s no exploding CPUs, no degrading CPUs and certainly don’t have to deal with dodgy power connectors.Yes, I'm sure it's Paradise on your side of the world.
The thing is, in Strix Point there is really little that could boost the FP performance. The only thing worth of note is updated scheduler layout, bigger register file and lower add latency. Where int side was boosted with improvements to load, store, and more execution units. I mean int scalar side has seen much more changes comparatively speaking.I don't understand the decision to beef up FP perf while barely moving the needle with int perf. Especially for Strix Point-
Zen 4 was more than capable in FP.
The main reason fp perf is suddenly important in laptops is the AI hype.
But you already have the NPU and GPU for those workloads.
You are right and AMD is already advertising total SoC AI TOPS on their product pages.The main reason fp perf is suddenly important in laptops is the AI hype.
But you already have the NPU and GPU for those workloads.
Only just found that out recently when it was mentioned here.
Cannot think of any valid reason any benchmark would pause between tests as that would just skew things in favour of bursty behaviour especially in a lightweight benchmark like GB.
Don't know about any Apple prejudice but it certainly favours any device whose cooling cannot cope with full usage.
Well, I guess GB6.4 should now come out with fanny and fanless scoresNo, that's a blatant proof that Geekbench was initially developed for embedded platforms.
Work smarter, not harder.Try running your precious silicon at 5.5 GHz.
Suggs likes his FPU/SIMD thicc and fast.I don't understand the decision to beef up FP perf while barely moving the needle with int perf. Especially for Strix Point-
Historically shortcoming for consumers, manufacturers love a bit of planned obsolescence.A historical shortcoming of LPDDR was that it only came in soldered form, but now that's changing thanks to LPCAMM.
We don't know which issues they solved by splitting them I guess they did not do that out of malice to the user. Also people mention that Turin-D is 16core CCX, but it is 16 times the same core, with the same V/F curve etc.If amd does 12-core APU in the next gen, they should not split them into such different CCXs. Too many issues.
Is it favoring a brand or the result of other brands getting optimization work earlier on and that other brand just getting its own optimations now.At GB they knew what they were doing, and i dont think that the goal was to help some obscure smartphones vendors that produce mediocre and cheap designs.
And i wont even talk of GB 6 numbers in respect of GB 5 when it comes to some given brands, look the same as Cinebench whose updates seems to always favour the same brand, just look at the evolution from R15 to R20/R23 and R2024.
We're back in business baby!OK!! The fabled 35% has finally reared its head! lol