Wednesday, August 24th 2022

AMD Ryzen 7 7700X "Zen 4" Cinebench R20 Score Leaked

The Cinebench R20 score of an AMD Ryzen 7 7700X "Zen 4" processor (possibly engineering sample), was allegedly leaked to the web by "Extreme Player Hall," a video-format tech news published on Bili Bili, as discovered by 9550pro on Twitter. The 8-core/16-thread processor was shown scoring 773 points in the single-thread test, and 7701 points in the multi-threaded one. These numbers put it 25-30 percent faster than the current Ryzen 7 5800X, as pointed out by Greymon55. The multi-threaded performance of this chip is roughly on par with that of the 5900X, which means AMD is overcoming a 50% CPU core-deficit on the backs of higher IPC and memory bandwidth.

The 25% single-core performance gain over the 5800X, if extrapolated to other less-parallized workloads such as gaming, could put this processor about 5-10% ahead of the 5800X3D, and about 4-9% ahead of the Core i9-12900K. The 7700X could face an uphill task measuring up to "Raptor Lake" in multi-threaded tests, given that Intel is doubling down on its Hybrid Architecture, with more E-cores across the lineup. AMD may still have a crack at matching Raptor Lake's gaming performance with future variants that have 3DV Cache.
Sources: Extreme Player (Bili Bili), HXL (Twitter)
Add your own comment

44 Comments on AMD Ryzen 7 7700X "Zen 4" Cinebench R20 Score Leaked

#26
Mussels
Freshwater Moderator
This is some pretty big gains

Intel supposedly has some too, but with the opposite cause - they added more E cores, instead of faster with less
Posted on Reply
#27
Minus Infinity
thebluebumblebee"with future variants that have 3DV Cache."

Ah, maybe that's why there's no "800" CPU. Maybe it will be the Ryzen 7 7800X3D???
v-cache is coming earlier than expected, late Q1 or early Q2 2023 and is already showing massive increases compared to Zen 3 5800X3D. Clocks will also be virtually the same anon v-cache models. 7900X3D would be my sweet spot, but so far only 7800X3D and 7950X3D confirmed.
Posted on Reply
#28
gruffi
phanbuey

Sweet! Enjoying my 7700X ahem 12600K here...
Those are no stock 12600K numbers. Which makes that comparison quite pointless. Your overclocked 12600K probably needs twice the power of 7700X. Your 12600K will look like inefficient garbage compared to stock 7700X. So, no. You don't really enjoy a 7700X. ;)
Posted on Reply
#29
phanbuey
gruffiThose are no stock 12600K numbers. Which makes that comparison quite pointless. Your overclocked 12600K probably needs twice the power of 7700X. Your 12600K will look like inefficient garbage compared to stock 7700X. So, no. You don't really enjoy a 7700X. ;)
Someone's salty lmfao. Actually it's quite efficient (and quiet). I do, and have been enjoying my efficient garbage for a while now :toast: .
Posted on Reply
#30
gruffi
ModEl4I was expecting more.
If it's indicative it will give something like the below in CB23 multithreading if they squeeze every ounce of 230W:

7950X 37130
7900X 29110
7700X 19820
7600X 15170

Intel Raptor should look something like the below:

i9 13900K 38940
i7 13700K 28540
i5 13600K 22890
i5 13400 16050

AMD just can't compete with the old pricing ($799-299) level!
Intel recently dropped a price warning of up to over 20%. RPL is a lot more expensive to produce than ADL. Intel will probably increase prices. They need to increase prices. Their client margins are really bad. From an economical point of view ADL is a failure. So, AMD will have no problem to compete with Zen 3 pricing. CB R23 alone also is irrelevant. It's a best case scenario for Intel. Average performance and efficiency will define pricing.
phanbueySomeone's salty lmfao. Actually it's quite efficient (and quiet).
No, it isn't. And I know that you are salty looking at Zen 4. Your comment proved that. ;)
Posted on Reply
#31
phanbuey
gruffiNo, it isn't. And I know that you are salty looking at Zen 4. Your comment proved that. ;)
:laugh: Oh yes, me being happy that my oc'd i5 performs even close to 5nm zen4 in any benchmark is definitely salty.

Im going to go cry now, you definitely defended AMD's honor good job. I'll let Lisa know you're the best AMD fan out there. Maybe you can get a sticker.
Posted on Reply
#32
1d10t
Was expecting more but if efficiency and platform longevity are guaranteed, I'm sold :D
Posted on Reply
#33
gruffi
gffermari….and Intel is already there. AMD will need to make the jump into it in Zen 5.
But AMD is supposed to take a different route that looks a lot more promising. Intel's big little design actually is quite bad. It's not worth to waste your time with. P and e cores are not fully ISA compatible. Which means a lot of hassle for developers and gimped features for consumers (e.g. missing SMT and AVX512 on the e cores). For some apps you even have to deactivate all the e cores to get the best results. People also reported about less responsive systems compared to 11th gen. AMD won't develop a separate e core. Zen 4c and 5c are based on the Zen 4 and 5 p cores. Just a lot more streamlined for better area and power efficiency. Which means the c cores still offer the same feature set as the p cores. Causing much less headaches for developers and consumers.
phanbuey:laugh: Oh yes, me being happy that my oc'd i5 performs even close to 5nm zen4 in any benchmark is definitely salty.

Im going to go cry now, you definitely defended AMD's honor good job. I'll let Lisa know you're the best AMD fan out there. Maybe you can get a sticker.
After you reported back to Intel's headquarter? Unlike you I'm not interested in any company. And unlike you I don't need to defend my system with pathetic comparisons and claims. Get a reality check. Your 12600K won't be on the same level as 7700X. Even though you tried hard to make it look like that. Poor try.
Posted on Reply
#34
ModEl4
gruffiIntel recently dropped a price warning of up to over 20%. RPL is a lot more expensive to produce than ADL. Intel will probably increase prices. They need to increase prices. Their client margins are really bad. From an economical point of view ADL is a failure. So, AMD will have no problem to compete with Zen 3 pricing. CB R23 alone also is irrelevant. It's a best case scenario for Intel. Average performance and efficiency will define pricing.
The up to 20% figure concerns both core and Xeon lines, imo Intel will not go above the below 1K unit prices:

13900KS $799 (+$60)
13900K $649 (+$60)
13700K $459 (+$50)
13600K $319 (+$30)
13400. $219 (+$27)

With these prices AMD just can't compete with the Zen3 era margins/SRPs.
Regarding Alder Lake's margins I won't even comment...
Maybe AMD will attempt to price them high, but probably within one month from Raptor's Lake launch there will be a correction regarding street prices.
We will see!
Posted on Reply
#35
Icon Charlie
Yea... This overall does not look that promising for AMD. This is the reason why there is no 3DV on the 5900 nor the 5950XT They do not want to have this generation of CPU's to overshadow the gaming performance of the newer Generation that is coming online.

As one person put it. "I can buy a quality Intel CPU and motherboard and be of equal performance and cost less than AMD has.

You are going to get screwed over by AMD's motherboard prices. Again from what I heard DOUBLE the price on most motherboards.

If that is true then I'll go the Intel Route.

I'm not a AMD fanboi. Nor an Intel one as well.

I go for the best bang for the buck. If AMD is going to screw me over on the cost of the mother board, I'll take my money, like many that will and buy something else.
Posted on Reply
#36
gffermari
The problem is many expect AMD to be cheap no matter what while Intel has always the right to be ok, expensive or ridiculously expensive.
So for them it’s always intel vs AMD with same performance but cheaper.
Posted on Reply
#37
Mussels
Freshwater Moderator
I'm all for quoting intel power figures at people, but avoid the fanboy hashing. We're due for months of it non-stop with both companies having some good looking tech this year, go easy.

@phanbuey I'm sure its powerful and not an issue at all for most tasks, but theres no way at 5.2GHz that it fits the definition of "efficient" which is what this seems to be about
The only thing seperating it from a 12900K is two P cores, so its power consumption simply isn't that much lower

Single threaded, or the stock 125W power limit and it's certainly efficient enough - but not with an all core OC, the moment the thread counts go up in applications that performance comes at a wattage cost that just destroys the efficiency.

TPU's review only did 4.9GHz and it added 100W over stock


And uhh... it didn't do so hot, this is the chart you want to be at the bottom of, not the top.



TPU did run at 1.45V to achieve this, so your specs claimed 1.235v should be significantly better - but i cant find reports of anyone else achieving that speed at less than 1.35v, with most needing 1.35v+ to even reach 5GHz.

If you do have a golden chip, that explains your love for it and efficiency claims - but anything other than that and you have one of the least energy efficient CPU's in existence :/

This is why people argue and it turns into fanboy wars.

If anyone is going to comment in a news thread about your superior hardware, post some screenshots and proof or it will only be interpreted as fanboyism or trolling. Show HWinfo64 with the wattages and clock speeds to backup the claims, otherwise we have to google and hunt down info and no one's gunna believe it.

They've got great ST performance, and they can have great MT performance - but it's a direct choice between that performance, or the power efficiency as it's not possible to have both at the same time.
Posted on Reply
#38
Mussels
Freshwater Moderator
AM4isGODJust like powerful cars, i would have thought PC power users hunt power not efficiency.

If you wanted an efficient rig you would not be a power user surely. I do get the wanting a powerful rig without the massive power use, but the high end parts, 3090/12900k/5950x are surely all high end parts so efficiency is not a priority.
No?
Why would anybody want that.
I get 50MHz less from my 3090 with 110W less power consumption by undervolting it.

I've got a 3090 and top tier gaming PC that doesn't even use 300W of power.


Why would I ever, ever consider making it use more power for no performance gains?


Heres 4K 140FPS gaming, DX12 - Unreal engine 4 specifically
No specs hidden or misleading information, it's a 3090 at 1.7GHz and a 5800x at 4.6GHz


And it uses 400W, with two monitors connected to that UPS.
GPU and CPU run under 40C and 60C, silently, always.


Am i meant to think my GPU wattage is a score? Do i need to crank this to the stock 375W to enjoy it?


Oh no, 6% more performance for 46% more wattage, how will i ever sleep at night

Especially with my 99% minimum FPS dropping... oh no, wait it's higher underclocked since the boost algorithm runs into TDP throttles. Nevermind.
It also gets worse as time goes on and the GPU heats up, this is custom loop with an active backplate with 15C ambients - they dont sustain those clocks at all on air, or in summer.
Posted on Reply
#39
AM4isGOD
MusselsNo?
Why would anybody want that.
I get 50MHz less from my 3090 with 110W less power consumption by undervolting it.

I've got a 3090 and top tier gaming PC that doesn't even use 300W of power.


Why would I ever, ever consider making it use more power for no performance gains?


Heres 4K 140FPS gaming, DX12 - Unreal engine 4 specifically
No specs hidden or misleading information, it's a 3090 at 1.7GHz and a 5800x at 4.6GHz


And it uses 400W, with two monitors connected to that UPS.
GPU and CPU run under 40C and 60C, silently, always.


Am i meant to think my GPU wattage is a score? Do i need to crank this to the stock 375W to enjoy it?
It's not about MAKING it use more, why would anyone deliberately do that.

You have undervolted your 3090, how many 3090 user you think have done the same. Some people throw money at their rig to have a high end rig to show off on their favourite forum without a care for how much it uses.

By hunting power i did not mean watts, but i guess you did not get my meaning, i meant performance. You really thought i meant power as in watts?

I get wanting as much performance as possible with as little power. My point was trying to be that most high end components are not really designed to be efficient, but i guess belittling my post was more important to you. Guess you ignored this-I do get the wanting a powerful rig without the massive power use.
Posted on Reply
#40
wolar
My main concern with this whole 3D cache is that they increased the price substantially for the expected improvement over time, thus raising the pricing once more for something that should be, well, expected.
Slowly but surely this turns out the same as nvidia did after the 9xx series, which raised the price to get the expected improvements.
Posted on Reply
#41
phanbuey
@Mussels That's not at all what's it's about. It's really about me teasing someone for crying over some benchmark results. No one is expecting a 10nm i5 to be more efficient than a 5nm brand new arch lol.

Totally off topic but here's some efficiency numbers for this sample:
Thread based boost is a lie | TechPowerUp Forums

By the way the difference between a 4.9Ghz uv and 5.2Ghz OC performance wise is not noticeable in real life. ADL, like your 3090, undervolts great and you lose very little performance per watt (I also undervolt my 3080 because that extra 50% power consumption for nothing makes no sense) :toast:
Posted on Reply
#42
gruffi
phanbuey@Mussels That's not at all what's it's about. It's really about me teasing someone for crying over some benchmark results. No one is expecting a 10nm i5 to be more efficient than a 5nm brand new arch lol.
Alder Lake was as "brand new" compared to Rocket Lake as Zen 4 is to Zen 3. Nanometers also don't say much. Intel's 10nm process is comparable to TSMC's 7nm process. Intel's 10nm is even slightly more dense. 7nm Zen 3 is still a lot more efficient than 10nm Golden Cove, area and energy wise. But to see that you must not compare Vermeer based SKUs but Cezanne based SKUs. Vermeer is somewhat handicapped by the old and power hungry Glofo 12nm IOD in single core scenarios. Cezanne is a monolithic design entirely manufactured on TSMC's 7nm process. NBC measured mobile Golden Cove with 22W and mobile Zen 3 with 14W single core, while the former one was about 10-15% faster. Golden Cove including L2 is ~7mm², while Zen 3 including L2 is ~4mm². So, it's not just the process node. It's Intel's p-core architecture itself that is a lot more inefficient than the competition. That's also one of the reasons why they are trailing so much behind in the server space.
Icon CharlieYea... This overall does not look that promising for AMD. This is the reason why there is no 3DV on the 5900 nor the 5950XT They do not want to have this generation of CPU's to overshadow the gaming performance of the newer Generation that is coming online.
That's wrong. The reason why there are no V-Cache versions of 5900X/5950X is Milan-X. 5900X3D/5950X3D would need two V-Cache chiplets. That's a lot of wasted chiplets for AMD. Because capacity is limited and Milan-X has much better margins. 5800X3D needs only one V-Cache chiplet and isn't really slower at gaming than 5900X3D/5950X3D would be.
Icon CharlieAs one person put it. "I can buy a quality Intel CPU and motherboard and be of equal performance and cost less than AMD has.

You are going to get screwed over by AMD's motherboard prices. Again from what I heard DOUBLE the price on most motherboards.
Who said that? Obviously he never bought an AMD system. Alder Lake boards are ridiculous expensive. The same will apply to Raptor Lake boards. That's no argument against Zen 4 and AM5. The best budget system is still AM4 based. New systems will always cost more. That's no news.
Posted on Reply
#43
Mussels
Freshwater Moderator
AM4isGODIt's not about MAKING it use more, why would anyone deliberately do that.

You have undervolted your 3090, how many 3090 user you think have done the same. Some people throw money at their rig to have a high end rig to show off on their favourite forum without a care for how much it uses.

By hunting power i did not mean watts, but i guess you did not get my meaning, i meant performance. You really thought i meant power as in watts?

I get wanting as much performance as possible with as little power. My point was trying to be that most high end components are not really designed to be efficient, but i guess belittling my post was more important to you. Guess you ignored this-I do get the wanting a powerful rig without the massive power use.
Considering all the massively popular undervolting videos on youtube and all the guides... a lot of people.
Overclocking is for low end hardware to get more performance. High end hardware has no headroom, you undervolt or throw power at nothing.
Posted on Reply
Add your own comment
Nov 27th, 2024 19:57 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts