I think the best estimates are that Zen4x3D will be 10-15% better than intel. Even if it just matches the best intel has that’s not too bad as there are at least a further two upgrades on the AM5 socket with Zen5 and Zen5X3D.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Poll: Ryzen 7950X3D, 7900X3D, 7800X3D
- Thread starter kindai
- Start date
I think the best estimates are that Zen4x3D will be 10-15% better than intel. Even if it just matches the best intel has that’s not too bad as there are at least a further two upgrades on the AM5 socket with Zen5 and Zen5X3D.
Associate
10-15% faster in gaming in CPU limited scenarios cause X3D does it do any benefit for anything except games?
Don’t think you will see a benefit as you have the standard Zen 4 cores for non gaming tasks. These should match up to known Zen 4 benchmarks for these programs.
Caporegime
Zen 4 is better for clocks than most people think, stock they clock up to 5.7/5.8Ghz on multiple cores simultaneously in games.
I don't think the 13900K is able to sustain 5.8Ghz on 8 cores out of the box, with the 8 or 9% higher IPC the advantage still lays with the 13900K but most reviewers have that between 5 to 10% overall, i think Intel does better in lighter threaded games and AMD slightly better, relatively in high threaded games as Intel's maximum boost is more single core.
The 5800X3D with the RTX 4090 is on average 23% faster than 5950X at 1080P, so i think around 15% better than the 13900K is not unreasonable tho i think 10% is a safer bet.
I don't think the 13900K is able to sustain 5.8Ghz on 8 cores out of the box, with the 8 or 9% higher IPC the advantage still lays with the 13900K but most reviewers have that between 5 to 10% overall, i think Intel does better in lighter threaded games and AMD slightly better, relatively in high threaded games as Intel's maximum boost is more single core.
The 5800X3D with the RTX 4090 is on average 23% faster than 5950X at 1080P, so i think around 15% better than the 13900K is not unreasonable tho i think 10% is a safer bet.
Last edited:
Soldato
Well, I've heard it said that something is stable for you if it's stable doing all the things that you want it to do. In Jay's case if all he wants it to do is run 20 seconds of Cinebench R23 every now and then, then it's stable for him...
It's possible that somewhere there are applications that wouldn't be stable on your setup, but since you've not run every possible thing in the world to test your system, you're happy that it's stable for what you want.
I'll be honest, I couldn't care less about the size of a core. If it's twice the size, costs he same and performs 10% better then I don't see the issue. If it run hotter then that would bother me, if it runs cooler, perhaps because of the size, then I'd see that as a posative.
That said, my concern with all the new CPUs from either side is the heat output.
Caporegime
@GoogalyMoogaly You clearly didn't read everything i said.
Associate
Yes only one CCD on both the 7950X and 7900X gets the 96MB 3D V Cache. The 78003DX gets 96MB just like 5800X3D.
I think 7900X3D and 7950X3D are both meant for productivity and also gaming without having to build separate PCs so the rig can have best of both worlds. So they make one CCD for gaming with 3D cache and other CCD a normal 7900X or 7950X one with only 32MB cache for assisting in productivity tasks cause like no games have any tangible benefit from more than 8 cores and very few even from more than 6 cores.
Though not sure how much lower clock speed of the CCD with the 3D cache will hold back productivity of the whole chip. though at least AMD is trying to give best of both worlds in one chip, though lets see how it goes. And now sure if normal CCD will have to downclock to speed of 3D cache CCD in all core workload (which games are not as no game uses 12 or 16 cores) or if they can run separate speeds in CInebench R23 and such?
Last edited:
Associate
Zen 4 is better for clocks than most people think, stock they clock up to 5.7/5.8Ghz on multiple cores simultaneously in games.
I don't think the 13900K is able to sustain 5.8Ghz on 8 cores out of the box, with the 8 or 9% higher IPC the advantage still lays with the 13900K but most reviewers have that between 5 to 10% overall, i think Intel does better in lighter threaded games and AMD slightly better, relatively in high threaded games as Intel's maximum boost is more single core.
The 5800X3D with the RTX 4090 is on average 23% faster than 5950X at 1080P, so i think around 15% better than the 13900K is not unreasonable tho i think 10% is a safer bet.
Well I can run all cores at 5.6GHz fully stable all workloads on air on a 13900KS with e-cores disabled and 5GHz ring clock. With AMD those clocks are dynamic and depend on workload and how often can it reach those clocks as they are not static all core all the time?
Soldato
did too
Caporegime
I don't have one, i can only observe what i see from others.
Taking my own CPU, which is a Ryzen 5800X (Zen 3) its advertised as "Up to" 4.7Ghz, at stock it runs at between 4.6 and 4.7Ghz on all cores simultaneously in a high stress workload, in games its locked solid pretty much to 4.85Ghz on all eight, unless those cores have very little or no load on them, in which case they can drop to as low as 3.7Ghz, a sort of power saving state.
Edit: bad grammar.
Last edited:
Soldato
Cache sizes
7950X3D - 144 MB (64 MB CCD, 64 MB V-Cache + 16 MB L2)
7900X3D - 140 MB (64 MB CCD, 64 MB V-Cache + 12 MB L2)
7800X3D - 104 MB (32 MB CCD, 64 MB V-Cache + 8 MB L2)
I think only few things outside of gaming the cache makes a difference
If anything might be slightly slower in productivity over the none 3d models because the CCD with cache are lower clocked
Last edited:
Caporegime
I don't normally run my OSD with all the core clocks displayed but i managed to find this one, its an early test map of mine, Unreal Engine 4.
Notice the cores that have any load on them are at 4.9Ghz, that's how it works dynamically, the opposite of how most people think it would. I had a Star Citizen screen shot where i had all 8 running at 5Ghz but can't find it... i don't run the CPU with that much extra boost anymore as its not as stable as i need it, i had the CPU for i think about 3 days and was still 'ing about with it. The UE screenshot is how i'm running it now 24/7.
Last edited:
Deleted member 258511
D
Deleted member 258511
Oh I know they are all the same. Does this happen with consoles or phones or TV’s even? Im sure they give a little more info before release. Im going way off topic here but they could give us something for **** sake.
Phones usually get detailed information leaked before release and TVs seem to get reviewed well before any release dates. So cpu and gpu manufacturers seem to be very secret squirrel.
Stock vs stock is fair. To unlock the "full potential of 13900k" you need the following (using OCUK prices as per forum rules)
z790 motherboard (£690 for Maximus Hero)
Contact frame (£45)
Custom watercooling (£300-500+)
8000Mhz DDR5 (last I saw in stock in UK was £580 for Gskill kit)
Much time to tune the LLC voltage for 5.8Ghz-6Ghz all core
Whereas with the 7950X3D, assuming it lives up to performance expectations:
Cheaper motherboard, £300 B650 will do
Cheap air cooler (£50 one will do)
6000Mhz DDR5 (£199 for 32GB)
Auto BIOS settings, 10 seconds to configure
Last edited:
Permabanned
- Joined
- 28 Sep 2018
- Posts
- 0
to unlock the full potential, you have to know what you’re doing and not be “dave2150”
Soldato
- Joined
- 14 Aug 2018
- Posts
- 3,558
Ah Dave, it doesn't seem to matter which side you're batting for, your ability for hyperbole remains undiminished!
There's just over two weeks to go so maybe let's wait until then before making these "but what about..." comparisons.
Last edited:
Soldato
I mean he’s sort of right.
Intel can be tuned to a much larger degree so if you are running stock you are indeed leaving performance on the table but you don’t have to do it.
Like the 5800X3D these new vcache models will be more plug and play and more suitable for mainstream as you are limited in what you can do.
2 Dimm board Z790I edge or Apex
contact frame
7200 kit should XMP without much trouble on a 2 dimm board but if want 7600 + and/or all timings optimised then need knowledge of DDR5, good IMC and testing = time, or money from a consultation
Decent cooler is fine for gaming 200W for headroom
custom loop direct die or delta TEC for every last % for silly money
Intel can be tuned to a much larger degree so if you are running stock you are indeed leaving performance on the table but you don’t have to do it.
Like the 5800X3D these new vcache models will be more plug and play and more suitable for mainstream as you are limited in what you can do.
2 Dimm board Z790I edge or Apex
contact frame
7200 kit should XMP without much trouble on a 2 dimm board but if want 7600 + and/or all timings optimised then need knowledge of DDR5, good IMC and testing = time, or money from a consultation
Decent cooler is fine for gaming 200W for headroom
custom loop direct die or delta TEC for every last % for silly money