No, AMD just likes doing CBT with their physical design engineers. Remember Navi3x had nearly double the frequency target from Navi1x with just a single node shrink.
There is a different amount of work done in some frames compared to others. Consider for example when particle effects litter the screen following an explosion.
And depending on what instructions are used that could either increase or decrease the clock rate since they're limited by thermals and voltage. Being able to clock absurdly high when there are 'low thermal impact' instructions in the mix means it will finish that frame faster.
I don't think there is much reason to be happy that AMD is chasing clock rates but performance is performance. If they can reach 6GHz it should be enough
Josh128
Golden Member
- Oct 14, 2022
- 1,051
- 1,587
- 106
EPYC Zen 5 comes nowhere near +40% vs EPYC Zen 4, in SPECint or SPECfp overall, so it doesnt matter.
6 GHz (or 5%) would not be enough, IMO, because of:
- couple of node jumps
- jumping to performance optimized version of N2
- the goal of Zen 6 is to clock optimize Zen 5 design
I think minimum of 10% should come out of this, which would be ~6.3 GHz.
No, 6 GHz is fine.
Anything beyond that is expecting too much. But that is the tradition for speculation threads.
poke01
Diamond Member
- Mar 8, 2022
- 3,733
- 5,077
- 106
6GHz isn’t fine. A mere .3GHz increase after 4 years would be bad and that too on N2P.
I agree that the clock increase should be a minimum 10%.
It is fine. Are you people even living in reality? The last generation for Intel regressed hundreds of MHz with a huge process improvement. AMD achieved nothing with a decently tuned process.
N2P won't do 10% for Zen 6 (6325MHz) in 1T unless AMD has wizards in its employ and why wouldn't Nvidia poach them? nT, sure.
Last edited:
poke01
Diamond Member
- Mar 8, 2022
- 3,733
- 5,077
- 106
Intel isn’t the basis of good cpu design anymore. Their P core are horrible. Anyway, yes AMD has good engineers. 6GHz isn’t some mythical barrier that can’t be broken.
How do anyone here determine what is a realistic speed?
Historical norms?
Physical fundamentals of the node?
This is how it has been and thus will always be, AKA faith?
Any room for a new approach?
These debates remind me of the time when IPC was relatively stagnant following the Haswell period, and was widely assumed to be the unchangeable future.
My take.
Historical norms?
Physical fundamentals of the node?
This is how it has been and thus will always be, AKA faith?
Any room for a new approach?
These debates remind me of the time when IPC was relatively stagnant following the Haswell period, and was widely assumed to be the unchangeable future.
My take.
FYI,
Zen 4 achieved 16% clock speed increase while going up only 1 node.
You are expecting Zen 6 to achieve 5% clocks speed increase while going up 2 nodes.
I didn't say that. But when it ships at a much lower clock rate than the same darn hypebeasts suggested don't claim anything about disappointment. 5% higher frequency combined with the earlier slide suggesting >10% IPC would be in line with past Zens.
The same few people are setting up another AMD disappointment train. They have fun doing the same thing over and over, I guess.
7GHz would improve the perception of Zen 6% a bit
It's not just the overall performance increase, but where it is going to come from.
- from core itself, probably close to 0%
- from bigger L3 slightly lower memory latency - ~5%
- from clock speed 10+%
So you are more optimistic on IPC, I am more optimistic on clock speed.
Why stop at 7Ghz? May as well go for 10Ghz or 100Ghz.
poke01
Diamond Member
- Mar 8, 2022
- 3,733
- 5,077
- 106
We will have to wait until Page 500 for the real answer.
In the meantime, I don't think we will have any trouble filling pages 192-499 with speculation.
poke01
Diamond Member
- Mar 8, 2022
- 3,733
- 5,077
- 106
Yes I got these numbers from my bottom of my ass just like MLID did.
What do you mean FYI? Of course I know. But why do you act like N3 increased peak frequency at all? It helped in other ways. Effectively Zen 6 will have only one node worth of peak frequency gain and after Zen 3 and Zen 4 achieved the low hanging fruits.
And where in Apple's voltage range would Zen 4/5 voltage @ peak freq be? Off the charts if I remember correctly.
That's more like where Turin is and lo, the gains were good even on 4nm.
gdansk's Theory of Rumors: Zen n rumors are likely to be directly in proportion to Zen n-1 rumors.
E.g. as Zen 5 achieved about 35% of the rumored IPC increase so too shall Zen 6 achieve about 35% of the "rumored" peak frequency increase.
And they're likely to be proportional because in both cases they're from the same sources. Extrapolating wildly from shreds of information. It will be justified after the fact by pointing to server products which achieve something close. I.e. Venice may gain ≥20% frequency in all core work loads but somehow this telephone game ends with a rumor of Zen 6 aiming for 7GHz.
E.g. as Zen 5 achieved about 35% of the rumored IPC increase so too shall Zen 6 achieve about 35% of the "rumored" peak frequency increase.
And they're likely to be proportional because in both cases they're from the same sources. Extrapolating wildly from shreds of information. It will be justified after the fact by pointing to server products which achieve something close. I.e. Venice may gain ≥20% frequency in all core work loads but somehow this telephone game ends with a rumor of Zen 6 aiming for 7GHz.
Last edited:
Win2012R2
Senior member
- Dec 5, 2024
- 977
- 977
- 96
Win2012R2
Senior member
- Dec 5, 2024
- 977
- 977
- 96
They get a bonus (2 weeks holiday) if they hit target freq level on all cores with passive cooling?
Physics? Power usage should go through the roof with that kind of freq.
MLID covers his arse with "target frequency" weasel words, total bait to get views and clicks, half of his videos are: "I predicted correctly this and that years ago, blah blah", never a word about mispredictions.
If AMD actually authorised feeding him any info (right or wrong), then they should really re-examine it - fake expectations are very damaging.
Last edited:
I think AMD will just sell Zen 5 for 6 and 8 core CPUs.
We will see if Nova lake is able to outperform Zen 5 in gaming...
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter