Info Overclocked 16 core Ryzen 3000 CPU scores 4346cb in Cinebench R15 and beats the $2000 18-core i9 9980xe!

[NightHawkRMX]

Last night, AMD announced new 6, 8, and 12 core CPUs based on their 7nm Ryzen 3000 architecture. Contrary to speculation, AMD did not disclose any information on a 16 core cpu.

AMD demonstrated their 12-core AMD Ryzen 9 3900x is using 2x8 core CCX design, so it isn't crazy to imagine AMD would make a 16 core CPU with all cores enabled..

However, shortly after AMD's keynote, Aussie youtuber TechYesCity uploaded a video detailing an inside look at an upcoming 16 core Ryzen 3000 cpu.

This engineering sample 16 core, when overclocked, managed 4.25ghz @1.572v using water cooling. The 16 core/32 thread cpu managed a 4346cb in Cinebench R15. This score creams the 18 core i9 7980XE and 9980XE CPUs at stock frequencies, even though these i9 CPUs cost nearly $2000.

CPU	Base frequency	Maximum frequency	Cinebench R15 score
Intel Core i9 9980XE	3.00GHZ	4.4GHZ	3646cb
Intel Core i9 7980XE	2.6GHZ	4.2GHZ	3404cb
AMD Ryzen 3000 Series 16 core	4.25GHZ	4.25GHZ	4346cb

All Core i9 stock frequency Cinebench results are from https://www.overclock3d.net/reviews/cpu_mainboard/intel_core_i9_9980xe_18_core_cpu_review/6

TechYesCity's Video:

View: https://www.youtube.com/watch?v=MkO4R10WNUM&

 

[COLGeek]

While I, too, am looking forward to the 3rd gen Ryzens, I would recommend we all have a bit of patience and wait for some independent, non-marketing, reviews and benchmarks before declaring success.

Too many claims in the past turned out to be pure hype. In this case, I hope that is not the case.

The proof shall soon be in the pudding.

 

[NightHawkRMX]

While i agree that most manufacturer benchmarks are generally misleading and you should wait for 3rd party reviews, Brian from TechYesCity did not explicitly state where this result was from and i can bet that it wasn't from AMD. All he said was something like "this came from the overclocking guys behind the scenes "

Besides, im not sure how you can fake a cinebench run. Sure it favors amd, mostly because it favors cores, however the 16 core ryzen still beat the 18 core i9.
Still, getting 1000pts over a 7980xe in cinebench while having 2 less cores and a simmilar clockspeed is impressive. Maybe the Increased IPC and cache is helping out here.

 

[13thmonkey]

If you treat IPC as being the widest effect of memory and cache calls, as well as pure horse power then it literally cannot be anything other that IPC.

I've seen several comments recently treating IPC as being something that is questionable. IPC x Clockspeed = Grunt, so if core speed is the same, then the only explanations for raw processing power come down to IPC and Core count, there's no maybe about it. IPC is like gearing attached to an engine, an engine running at 4500rpm will give a greater speed with a more highly geared gearbox.

Assuming it's all true of course, and I also wish it to be true.

 

[RobCrezz]

Thats a pretty big overclock if it needed that voltage....

 

[Phaaze88]

Thats a pretty big overclock if it needed that voltage....

Something most users wouldn't achieve normally.
Give us scores with a more realistic overclock!

 

[Phaaze88]

Somewhat thread related...
So just about everyone knows the AM4 boards are compatible with all current, and soon to come Ryzens, but what of the lower end mobos?
Can the cheaper A or B series actually handle, say, if a user wanted to drop a 3900x in? I'm comparing this to the motherboard vrm situation with the FX 8000/9000 cpus.

 

[NightHawkRMX]

Thats a pretty big overclock if it needed that voltage....

Well, that overclock was on water cooling, not LN2.
The image with CPUZ at 4.1ghz 1.42v is what the guy on youtube stated is a more common overclock most users would do.
Not sure what voltages are normal for new ryzen yet.

 

[justin.m.beauvais]

Somewhat thread related...
So just about everyone knows the AM4 boards are compatible with all current, and soon to come Ryzens, but what of the lower end mobos?
Can the cheaper A or B series actually handle, say, if a user wanted to drop a 3900x in? I'm comparing this to the motherboard vrm situation with the FX 8000/9000 cpus.

The 3900X is a 105W CPU, which is the same rating as the 2700X. They should draw broadly similar amounts of power. There shouldn't be a repeat of the FX "crank it till it screams" power and heat problems.

 

[NightHawkRMX]

Well considering I have seen many 105w 2700x running full turbo in lower end B350 boards, I would assume the 105w 3900x could as well.
A320 won't get 3rd gen support. B350, B450, X370, and X470 will.
X570 doesn't support 1st gen, so that leaves b450 and X470 as supporting all 3 gens and having the most chance of running high-end ryzen cpus.

 

[13thmonkey]

There is a table on anandtech or tom's of what boards can possibly take what generation, I recall the A's can't take 3000's. But there is also manufacturer discretion. But as noted above rated power shouldn't be an issue, however longevity and size of boost may differ between the best and worst boards.

 

[NightHawkRMX]

This is what I remember.
If I recall, A320 will support Ryzen 1 and 2 with a bios update.
B350,B450,X370,and X470 can run 1st, 2nd, and 3rd gen with bios update.
X570 and maybe B550 can only run 2nd and 3rd gen ryzen

 

[drea.drechsler]

Somewhat thread related...
So just about everyone knows the AM4 boards are compatible with all current, and soon to come Ryzens, but what of the lower end mobos?
Can the cheaper A or B series actually handle, say, if a user wanted to drop a 3900x in? I'm comparing this to the motherboard vrm situation with the FX 8000/9000 cpus.

Just because it's compatible doesn't mean all the motherboard mfr's are going to release bios updates to support R3K on all their motherboards. That's already been discussed a lot.

Assuming you've a board that gets a bios update you probably could pop in a 3900X but I can't imagine seeing near the same performance under boost as a top-tier board with more phases and cool running FET's under a massively finned heatsink. That's even if you don't overclock -- which we've yet to learn anything about on these CPU's -- because the boosting profiles are usually established in BIOS by the motherboard mfr. based on the power delivery capability of the motherboard.

 

[iiSt0rmy]

Intel i9 9900k has a clock speed of 5.0GHz and is only $599 in my country

 

[justin.m.beauvais]

Assuming you've a board that gets a bios update you probably could pop in a 3900X but I can't imagine seeing near the same performance under boost as a top-tier board with more phases and cool running FET's under a massively finned heatsink. That's even if you don't overclock -- which we've yet to learn anything about on these CPU's -- because the boosting profiles are usually established in BIOS by the motherboard mfr. based on the power delivery capability of the motherboard.

Actually the performance increases on higher end boards come mostly from XFR. AMD rates their TDP and power usage stats at full CPU power draw at stock settings, very unlike Intel who rates their power and TDP at base clocks. The idea that VRMs are going to be a power bottleneck even on a properly rated board is a very Intel notion. So even with "cheap" VRMs all you are giving up is possibly some XFR and possibly overclocking. Running stock settings, a 3900X should run just fine on a BIOS updated budget B350 board, provided you don't plan to overclock much.

I'm shocked that people are automatically thinking that the 3900X is going to draw a LOT more power than the 2700X. It isn't. AMD has touted the same performance at reduced clocks and much reduced power thanks to the shrink to 7nm. We saw an example of it in Vega VII. If they had kept the same clocks performance would have been a little better with a substantial drop in power used. So, they cranked the clocks to get a better performing card in the same power usage range as Vega 64. In the Vega VII release they clearly explained what they did, and it should have gotten people excited about 7nm... but for some reason most people were just disappointed with 2080 levels of performance and didn't see that Vega VII was as much a tech demo of 7nm as it was a performance graphics card.

People are way too used to Intel just revising, increasing clocks, and releasing broadly the same architecture on the same process node. AMD isn't doing that.

 

[NightHawkRMX]

Well, I just got my bios updated to support 3rd gen, but my 4+2 phase VRM isn't going to be running an overclocked 12 or16 core anytime soon.
Let us see what the 3600 or 3600x are like.
https://wccftech.com/amd-ryzen-5-36...k-leak-crushes-coffee-lake-price-performance/ Seems I7 8700 and 3600 are close, with the 3600 winning in PUBG and geekbench. I'm wandering why amd and others are using this title to beat Intel. Tests were done with a 2080ti and 3000mhz DDR4.

 

[RobCrezz]

Well, I just got my bios updated to support 3rd gen, but my 4+2 phase VRM isn't going to be running an overclocked 12 or16 core anytime soon.
Let us see what the 3600 or 3600x are like.
https://wccftech.com/amd-ryzen-5-36...k-leak-crushes-coffee-lake-price-performance/ Seems I7 8700 and 3600 are close, with the 3600 winning in PUBG and geekbench. I'm wandering why amd and others are using this title to beat Intel. Tests were done with a 2080ti and 3000mhz DDR4.

Impressive if true. look forward to seeing real benchmarks.