- Mar 3, 2017
- 1,747
- 6,598
- 136
For western English reviewers I would watch Gamers Nexus and Hardware Unboxed. I crossed out Linus ages ago for reviews.Which ones do you have in mind?
I bet something like thisI'm still wondering how will the reviews present the - currently expected - ~16% ST gain compared to ~29% of Zen 4 or ~25% of Zen 3.
15% uplift is a whole lot better than a frickin' regression.
I actually think Zen 5 has bugs but still ended up exactly what Amd had envisioned. It doesn't seem impressive because Zen 4 had to do compete with Raptor Lake w/ sky high clock.uhm, I don't think AMD expect zen4 home users to upgrade to zen5 for ST performance.
If you're a gamer with zen3 or older zen5 might be a viable option or wait for the X3D version.
If you're a gamer with a regular zen4 you'll at least wait for zen5 X3D.
If you're a gamer with a zen4 X3D you'll wait for zen6
zen5 is an evolutionary generation that for home useres will help AMD stay competitive against intel, but mostly it will help them stay competitive in the Threadripper/EPYC segment.
And the CPU is far from launch. It is best to reserve judgement until the reviews are in. Then we can shame and blame AMD if needed
That's literally thumbnails for MLID videosI bet something like this
Total domination, Intel doomed, Intel should be worried... etc.
On Intel's launch,
Total Domination, AMDoomed etc..
Intel could've pushed them a bit to do better if only Intel had put Redwood Cove on Intel 7 and clocked it to 6 GHz.Indeed, but it is still disappointing considering previous progresses.
Thats really dependent on what exactly do you need to do, but for standard exterior archviz renders something like Octane Render is perfectly fine (aside of possible VRAM issue). When you put its output next to same picture done in something like Vray or Corona there is no visible downgrade in quality.The quality of the GPU render is awful next to the CPU render, and it's barely touched on in that blog post, and only in passing. The time comparison was definitely not to achieve the same end result.
I would do no different, would i have acquired the same level of fame like Igor just did. :-DAgain, you are lecturing me, as if I crafted the message. True or false, it does not change what AMD said. Their marketing message is "fastest consumer CPU in the world", this is how they positioned their product.
This one is GB 6.2.2 the other a few days ago was 6.3. Can’t these people use the latest Geekbench? 😅Highest scoring ST run so far, 4+6 ES doing
2839 @ 4.4, so around 645 pts/GHz.
@DisEnchantment you might be bang on the money with your assessment.
Acer Swift SF14-61 - Geekbench
Benchmark results for an Acer Swift SF14-61 with an AMD Eng Sample: 100-000000994-38_Y processor.browser.geekbench.com
Interesting, if this Lunar Lake result is representative, then the geekbench score per mhz is very-very similar (for the mobile parts)
Is the cache per core more for Zee5?Lunar lake has 8+4+12+8=32 MB of total cache, while Zen5 Strix has 36MB of total cache.
Rather, GB test loads are historically so short that CPU does not reliably boost to the max as the load stops before some cutoff is reached. It's not that in 'normal' (tm) tasks you have 10 ms long cpu loads and so on...OR the frequency fluctuation on Zen 4 is pretty random?
Also Geekbench famously has huge spreads in scores of the same CPUs based on software, OS and RAM config, so there is extreme margin of error when you just take two individual results and compare them.Folks attempt to make iso-core-clock performance comparisons based on benchmark runs on dynamically clocked computers. This approach has... caveats.
Another thing - the scores that i've published yesterday were done with CPU correctly boosting up to 5.75ghz, while GB shows (.gb6) that it was running at 5.5 ghz (which is actually the light multithreaded clock limit since some early-time AGESA). I checked it with hwinfo and it is congruent with dom's +200 fmax scoreAlso Geekbench famously has huge spreads in scores of the same CPUs based on software
It is weird that Geekbench often read the clock lower than spec when comes to new hardware which are not released.
Lunarlake U7 268V has 5.0Ghz max but GB read it 4.8Ghz max, it seems GB may need a revise or update the algorithm for these new hardware.
But it could explain those old Strixpoint GB leak which were also being read lower clock. These new CPUs are either poorly boosted in GB but still being read the clock correctly, or completely mess up that both clock reading and real boost clock are wrong.