- Mar 3, 2017
- 1,777
- 6,791
- 136
I like that trend very much. If AMD can keep improving the cores with higher single tread performance.
I was thinking server, but I guess you already answered that above.
adroc_thurston
Diamond Member
- Jul 2, 2023
- 7,071
- 9,817
- 106
They have to; the market is still very much moving.
Turin and Venice are both moar corez so yeah.
Single thread performance for non gaming workload doesn't matter much for almost all the client applications. Basically what I am saying is, for client applications single threaded performance is not a bottleneck. For tasks like rendering, having double the cores will have almost double the performance, which is much more than mere 30% for high IPC 16 core Zen5 Design. If rendering can't support 32 cores, you can run 2 instances of rendering. So a hypothetical 32 core Zen4 would perform better than 16 core Zen5 in multi threaded work loads.
Initially I didn't like the Intel P/E core design mainly because of huge difference in performance and ISA support. However for AMD if they can add 8core Zen5+16core Zen5C, that would be great and it would likely to perform better than 16Core Zen5 core where it matters. Unlike Intel, for AMD it is easy to switch CCDs because of chip-let design.
For me another generation of 16 core design is a huge disappointment. Space wise AMD can squeeze 3 CCDs in AM5 socket, but I would guess then they also need to update the IO die.
Initially I didn't like the Intel P/E core design mainly because of huge difference in performance and ISA support. However for AMD if they can add 8core Zen5+16core Zen5C, that would be great and it would likely to perform better than 16Core Zen5 core where it matters. Unlike Intel, for AMD it is easy to switch CCDs because of chip-let design.
For me another generation of 16 core design is a huge disappointment. Space wise AMD can squeeze 3 CCDs in AM5 socket, but I would guess then they also need to update the IO die.
adroc_thurston
Diamond Member
- Jul 2, 2023
- 7,071
- 9,817
- 106
What are the non-gaming applications that are bottleneck by single threaded performance? Who are actually suffering from it? How many people are saying "I wish if I could have little more single threaded performance" ?
I am not saying AMD shouldn't increase single threaded performance, what I am saying is they also need to increase the core count. Otherwise it will be just another 16 core chip.
JavaScript engines - web equals JavaScript. Web sites grow more and more complex due horrible bloat of JS frameworks, also Electron apps... Using those on a mobile Skylake is painful.
So yes, if you plan to browse the web in upcoming 5 years, you surely need MOAR single-thread.
coercitiv
Diamond Member
- Jan 24, 2014
- 7,354
- 17,423
- 136
You got your personal priorities mixed up with mainstream consumer priorities. For the consumer base, core count beyond 16 is not even a remote priority. Today consumers need faster ST and lower cost while hitting 8+ core count in the process.
Anyone who is seriously in need for more than 16 fast cores is a pro-sumer or professional, and their needs are not really representative of the mainstream consumer market. The only reason we see opinions like yours pop up every now and then is because both Intel and AMD have neglected the HEDT segment, either through attrition or prohibitive pricing. However, feeling unrepresented in their product offering does not make you an exponent for mainstream consumer needs.
I didn't find any web pages running slowly in 2700x or current 5950x to a point to say I need more single thread performance. 30% performance won't help you much with JS bloat-ness. I am happy with threaded performance increase, just that AMD also need to increase the core count. Otherwise it will be just like Intel continuing with 4 cores.
Last edited:
DrMrLordX
Lifer
- Apr 27, 2000
- 22,901
- 12,967
- 136
Yeah sure, that's kinda niche for the product in question though no?
Maybe Mark has some old hardware he could send your way?
I don't know much about Pixar Renderman, Blender Cycles or other examples, but for Autodesk Arnold at least they have basically kept pace with socket/core density as far as I know which is impressive given how fast that increased in the last 6 years.
OMG. When will this +16 Zen5c chiplet speculation die. It's a meme, it's not real.
Yes, the current IOD is dual channel. You'd need an extra channel for a third CCD or face bandwidth limited performance. But, it's cheaper to stick with the same IOD and the same chiplet formula for now. Redoing the routing layout for a 3 CCD with an extra channel would be a big challenge given the current package size for AM5. Going from Zen3 to Zen5 will pretty much be the equivalent of adding an extra CCD anyway - given the cumulative performance boosts from Zen3->Zen4 and then Zen 4->Zen5.
adroc_thurston
Diamond Member
- Jul 2, 2023
- 7,071
- 9,817
- 106
It will be a slightly souped up mobile chip.
adroc_thurston
Diamond Member
- Jul 2, 2023
- 7,071
- 9,817
- 106
No, much the same way -halo mobile isn't b.L.
You can pay for real cores.
Probably after Zen5 is EOL even if Lisa Su herself says it's never going to happen given that AMD PR is sometimes less than trustworthy Jebait tactics 😂
But I want my 96 core 144 thread 15 watt zen5 desktop cpu with 256mbx3d cache per chiplet that is Cinebench optimized for $199.99
Lol that would have to be running at sub 2 ghz at least, probably more like sub 1.5 ghz, and even then it's a real stretch.
The bell curve of optimisation only goes so far before dropping voltage gives diminishing returns of perf/watt, just the same as for the top end of processor perf where the power consumption starts rocketing up for minimal clock frequnecy improvements.
For 96 wider cores it's just not feasible.
Sarcasm is good but I see absolutely no relation between this comment and the possible consumer use of zen*C chiplets, which cost basically the same to AMD as vanilla chiplets.
So far the only, hypothetical, problem I read in this thread is the insufficient IOD-chiplet and/or DRAM-IOD bandwidth, with no definive arguments supporting this point.
Same market as the 7950X.
Gamers buy the 7800X3D, not the 7950X.
Go back and read whole thread. The dreams are large. I was joking. Also, AMD stated only 16 core for Zen 5 already I believe so that is moot. If they hit 20% plus single thread boost I cant see the need for larger than 16 core except for halo over Intel. Shrugs. I just want it release tomorrow as I have $5K for new pc ready for it.
Correct (well and the 7950X3D)... but the 7950X came out first. Time to market matters.
Correct, but the pricing for the original 7000 processors wasn't great to begin with. It wouldn't have been a massive issue if Intel hadn't launched their fervid processors a week later!
It's a sticky situation Had AMD released months prior they would have had trouble moving because of ddr5 pricing. had they released much later they would have still had to reduce pricing because of Intel being hot on their tail in performance and heat. DDR5 pricing would have been better.
now with zen 5 if they release early to mid q2 they have 2 quarters before Intel's arrow in the head with a lake of blood pooling processor comes out.
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 + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 384
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter