- Mar 3, 2017
- 1,687
- 6,243
- 136
Don't worry my people will be in touch with yours.38.4 Billion is small? (with 5.3% YOY growth?)
If you think that way can you send me a few million?
Yet you don't need to figure out what clock Zen5C is.View attachment 101012
15% faster in CB2024 at 45watts than M3 Pro 12 core (6P+6E).
This is surprising as half the of cores in the M3 Pro are LITTLE cores.
I wonder what the Zen5c cores are clocked at?
Ryzen 9 7950X (45 W) | 6544 |
Ryzen 9 7900X (45 W) | 6026 |
Ryzen 5 7600X (45 W) | 4985 |
Yet you don't need to figure out what clock Zen5C is.
I have gave up normalizing these Zen5/C cores and comparing to Zen4 Phoenix. I could only say the clock must be darn low when they're under stress load while TDP is limited.
Computerbase already did MT efficiency test.
AMD Ryzen 7000 im Test: So schnell sind 7950X und 7700X: Benchmarks in Apps und Games
Ryzen 9 7950X & Ryzen 7 7700X im Test: Benchmarks in Apps und Games / Leistung in Single-Core-Lasten (klassisch)www.computerbase.de
Cinebench R20 MT:
Ryzen 9 7950X (45 W) 6544 Ryzen 9 7900X (45 W) 6026 Ryzen 5 7600X (45 W) 4985
As you can see 7900x being 2x core count compare to 7600x, but under 45w it's only 1.2x faster, clock would be ~40% slower than 7600x or something. It would not be so different when comes to other platform like Strixpoint.
Great! Just a few will let me retire well!Don't worry my people will be in touch with yours.
Its smallish
Alot of people are one track minded gamers and don't see outside of that narrow lane. The industry is much bigger than just PC games.
I think there is plenty of people that will purchase Vanilla Zen 5 and of course gamers will go X3D.Great! Just a few will let me retire well!
If your point was simply games is just a small part of the market Zen 5 needs to address then I will 100% agree!
zen2 to zen3 had 200Mhz bump which is also quite small.It would have been nice if Zen 5 client was fabbed on N3E. Finflex is amazing. Then the clocks would have increased as well + 16% IPC would have been very nice.
Is this the first time AMD had no change in max clocks gen on gen?
zen2 to zen3 had 200Mhz bump which is also quite small.
200 MHz is small by today's standards where we have cores clocking well in excess of 5 GHz, but going from 4.7 GHz to 4.9 GHz for Zen 2 to Zen 3 was a 4.2% uplift. Not bad at all.zen2 to zen3 had 200Mhz bump which is also quite small.
What was the reason for that?Zen 2 couldn’t really hit its advertised boost clocks in real life usage whereas Zen 3 could easily sustain, and many times even exceed, its advertised boost clock.
N7 process improvements, i guess, late Zen2 cpus (which were later separated into the XT SKUs) clocked much higher than the earlier Zen2 samples (my CPU that was bought at release date could barely sustain 4.5ghz/4.3 ghz OC with custom loop, while XTs could easily do 4.5 ghz all-core at much more modest voltages). This same frequency improvement didn't happen with Zen3 (on the contrary, my december 2020 5900xt had both good chiplets and could do 4.8 ghz all core at very low voltage which not that many 5900x have beaten later) or Zen4 (release-time x3ds seem to have two top-notch chiplets while the next batch CCD1 was dogshit (mine can't even do anything above 5.75 ghz for ST boost, insta reboot etc).What was the reason for that?
What was the reason for that?
Binning is just hard.I think it basically came down to AMD still perfecting their boosting algorithm and being a bit too ambitious with the boost clocks.
One thing though, they would have going for Z6 is increased clocks & efficiency, of course assuming they would go for N3P, in 2026.Zen 6 having a ~10% IPC gain in 2026 would be a continuation of that trend.
Let me estimate two figures for the year 2026:Gaming "close" to X3D gains, [...]
They took almost 2 years from Z4 for these gains is the problem.
If Z6 takes 2 years for 10% then someone's lunch is going to get slowly shared with everyone
AMD does >10% when they launch X3D versions. Intel does about +10% per gen generally as well.Let me estimate two figures for the year 2026:
(Same as in 2024.)
- number of CPU makers who cater to the PC gaming market: 2 (two)
- number of CPU makers to whom "gaming IPC" increases of >10%/2y are a mission-critical design goal: 0 (null)
Great. So we're back to the dark ages, like when Intel were feeding us 6% CPU perf increase per year? If so, who should save us?number of CPU makers to whom "gaming IPC" increases of >10%/2y are a mission-critical design goal: 0 (null)
or in more clear terms, PPT is 230 W (default of the top Zen 4 SKU; let's see about Zen 5). Which is insane, as a ~32c/64t quadchannel workstation CPU would be a better fit into such a power envelope, IMO.AM5 platform TDP is 170W.
Vector throughput will easily make good use of this power budget obviously,¹ but this doesn't seem so important for most desktop usage scenarios. Non-vectorized multithreaded loads should see quite high sustained clocks with this much power available…What do we think AMD will use the remaining TDP headroom for?
Doing the outlook to 2026, I went with the 10% (and prepended a ">") from @DisEnchantment's reference to Zen 6, which presumably will have "leveraged" cores, IOW will supposedly be a "tick".But is it really that bad? Even if we'll only get 16% IPC increase with Zen5, it's better than what you mentioned.
Agreed. Which is why I’m still hoping for a moar cores mid-life kicker, despite that the gaming crowd in this thread does not care about such CPU variants.or in more clear terms, PPT is 230 W (default of the top Zen 4 SKU; let's see about Zen 5). Which is insane, as a ~32c/64t quadchannel workstation CPU would be a better fit into such a power envelope, IMO.
Great. So we're back to the dark ages, like when Intel were feeding us 6% CPU perf increase per year? If so, who should save us?
But is it really that bad? Even if we'll only get 16% IPC increase with Zen5, it's better than what you mentioned.
If you look at the multitude of scores it is by no means a guaranteed gain at all, let alone that good.AMD does >10% when they launch X3D versions