Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

Ok, this is Epyc (EPYC 7371 detailed, 3.4GHz on all 16)

Nicnac
 Share

AMD-EPYC.jpg

Here are the specs:

  • 16c/32t
  • 64mb of L3
  • 3.1 GHz base, 3.6 GHz (16 core) and 3.8 GHz (8 core) boost
  • 8 channel memory
  • 128 PCIe Lanes

 

This is AMD's highest ever clocked Zen CPU now with an impressive 3.6 GHz an all 16 cores!

With the same core count as TR how does it compare?

 

Quote

Now if we compare the chip to the Ryzen Threadripper 2950X part, another 16 core configured chip which has virtually the same core specs as the EPYC 7371, we see much higher clocks at 3.5 GHz base and 4.4 GHz boost. The part isn’t a fair comparison though since AMD is using their slightly optimized Zen+ cores. So if we compare it to the same Zen core based Ryzen Threadripper 1950X, we still get higher clock speeds of 3.4 GHz base and 4 GHz boost.

Looks like one hell of a CPU. Excited for another TR update maybe? 

 

 

From: https://wccftech.com/amd-epyc-7671-fastest-clocked-epyc-cpu/

 

My Folding Stats

 

X  Vigilo Confido  X

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Nicnac said:

This is AMD's highest ever clocked Zen CPU now with an impressive 3.4 GHz an all 16 cores!

You mean 3.6 on all not 3.4?

Link to comment
Share on other sites

Link to post
Share on other sites

Interesting to see a boost algo based on presumably CCX or perhaps just per chiplet (if still 2 CCX). Will they do a single core boost though?

 

Edit: Oh this is zen 1? Boring :(

MOAR COARS: 5GHz "Confirmed" Black Edition™ The Build
AMD 5950X 4.7/4.6GHz All Core Dynamic OC + 1900MHz FCLK | 5GHz+ PBO | ASUS X570 Dark Hero | 32 GB 3800MHz 14-15-15-30-48-1T GDM 8GBx4 |  PowerColor AMD Radeon 6900 XT Liquid Devil @ 2700MHz Core + 2130MHz Mem | 2x 480mm Rad | 8x Blacknoise Noiseblocker NB-eLoop B12-PS Black Edition 120mm PWM | Thermaltake Core P5 TG Ti + Additional 3D Printed Rad Mount

 

Link to comment
Share on other sites

Link to post
Share on other sites

this should help amd compete in more workload types, this 16 core still has the same L3 as the 32 core, which should be helpful, 3.8 on 8 cores is excellent in my eyes, as it will probably mean that most of the time it will be actually running at 3.8

Link to comment
Share on other sites

Link to post
Share on other sites

That's a good addition though, nice frequency increase. But yes 7nm soon to come, can't wait to see that specially consumer side. 

Ryzen 7 3800X | X570 Aorus Elite | G.Skill 16GB 3200MHz C16 | Radeon RX 5700 XT | Samsung 850 PRO 256GB |Mousepad: Skypad 3.0 XL | Mouse: Zowie S1-C |Keyboard: Corsair K63 MX red | OS: Windows 11

Link to comment
Share on other sites

Link to post
Share on other sites

I wish they'd also make CPU's that can clock stupid high when few cores are used and moderate when all of them are used. It seems only Intel does that, AMD is still rather conservative on top end clocks. Zen+ improved it (R7 2700X for example), but not quite enough for my liking.

AMD Ryzen 7 5800X | ASUS Strix X570-E | G.Skill 32GB 3733MHz CL16 | PALIT RTX 3080 10GB GamingPro | Samsung 850 Pro 2TB | Seagate Barracuda 8TB | Sound Blaster AE-9 MUSES Edition | Altec Lansing MX5021 Nichicon/MUSES Edition

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, RejZoR said:

I wish they'd also make CPU's that can clock stupid high when few cores are used and moderate when all of them are used. It seems only Intel does that, AMD is still rather conservative on top end clocks. Zen+ improved it (R7 2700X for example), but not quite enough for my liking.

It's not like they don't want to, they aren't being conservative, it's just that can't. More I'd that a limitation of the zen architecture or of the glofo processes used.

My Folding Stats - Join the fight against COVID-19 with FOLDING! - If someone has helped you out on the forum don't forget to give them a reaction to say thank you!

 

The only true wisdom is in knowing you know nothing. - Socrates
 

Please put as much effort into your question as you expect me to put into answering it. 

 

  • CPU
    Ryzen 7 1700 3GHz 8-Core Processor @ 4Ghz
  • Motherboard
    GA-AX370-GAMING 5
  • RAM
    DOMINATOR Platinum 16GB (2 x 8GB) @ 3400mhz
  • GPU
    Aorus GTX 1080 Waterforce
  • Case
    Cooler Master - MasterCase H500P
  • Storage
    Western Digital Black 250GB, Seagate BarraCuda 1TB x2
  • PSU
    EVGA Supernova 1000w 
  • Display(s)
    BenQ - XL2430(144hz), Dell 24" portrait
  • Cooling
    MasterLiquid Lite 240
Link to comment
Share on other sites

Link to post
Share on other sites

Maybe not on EPYC, but for Ryzen or Threadripper, they could use the bigLITTLE approach from ARM using recently hyped chiplets.

 

2-4 high performance, high clock cores and then bunch of regular ones for regular and heavy multithreaded tasks. And then scale the usage depending on workloads. Should be cheaper binning tiny chips high than entire chips.

AMD Ryzen 7 5800X | ASUS Strix X570-E | G.Skill 32GB 3733MHz CL16 | PALIT RTX 3080 10GB GamingPro | Samsung 850 Pro 2TB | Seagate Barracuda 8TB | Sound Blaster AE-9 MUSES Edition | Altec Lansing MX5021 Nichicon/MUSES Edition

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, RejZoR said:

Maybe not on EPYC, but for Ryzen or Threadripper, they could use the bigLITTLE approach from ARM using recently hyped chiplets.

 

2-4 high performance, high clock cores and then bunch of regular ones for regular and heavy multithreaded tasks. And then scale the usage depending on workloads. Should be cheaper binning tiny chips high than entire chips.

There's a recent video by AdoredTV where he talks a lot about binning and the impact 7nm has on it. Quite interesting

 

 

 

My Folding Stats

 

X  Vigilo Confido  X

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

gamer headset: on

knuckles: cracked

intel: crushed

 

 

yep, it's gamer time.

ASUS X470-PRO • R7 1700 4GHz • Corsair H110i GT P/P • 2x MSI RX 480 8G • Corsair DP 2x8 @3466 • EVGA 750 G2 • Corsair 730T • Crucial MX500 250GB • WD 4TB

Link to comment
Share on other sites

Link to post
Share on other sites

24 minutes ago, RejZoR said:

Maybe not on EPYC, but for Ryzen or Threadripper, they could use the bigLITTLE approach from ARM using recently hyped chiplets.

 

2-4 high performance, high clock cores and then bunch of regular ones for regular and heavy multithreaded tasks. And then scale the usage depending on workloads. Should be cheaper binning tiny chips high than entire chips.

You what? The 8 core chiplets are already smaller than a snapdragon 845. The entire point of AMD's plan is they just use those chiplets on everything. But yes the yields on those tiny chiplets is great

MOAR COARS: 5GHz "Confirmed" Black Edition™ The Build
AMD 5950X 4.7/4.6GHz All Core Dynamic OC + 1900MHz FCLK | 5GHz+ PBO | ASUS X570 Dark Hero | 32 GB 3800MHz 14-15-15-30-48-1T GDM 8GBx4 |  PowerColor AMD Radeon 6900 XT Liquid Devil @ 2700MHz Core + 2130MHz Mem | 2x 480mm Rad | 8x Blacknoise Noiseblocker NB-eLoop B12-PS Black Edition 120mm PWM | Thermaltake Core P5 TG Ti + Additional 3D Printed Rad Mount

 

Link to comment
Share on other sites

Link to post
Share on other sites

52 minutes ago, RejZoR said:

Maybe not on EPYC, but for Ryzen or Threadripper, they could use the bigLITTLE approach from ARM using recently hyped chiplets.

 

2-4 high performance, high clock cores and then bunch of regular ones for regular and heavy multithreaded tasks. And then scale the usage depending on workloads. Should be cheaper binning tiny chips high than entire chips.

Don't think it's worth it for anything but laptops. CPU cores are already small and the R&D required to make a competitive low power core is massive. Even for current laptops it's a bit of a stretch. You'd need to come up with a very good reason not just to keep the current paradigm or even just implement the same cores but on a different voltage plane and perhaps even with smaller cache. However even that requires extensive R&D. ARM had to do the same with DynamIQ in order to go from multi cluster to single cluster with mixed core and cache structure.

27 minutes ago, S w a t s o n said:

You what? The 8 core chiplets are already smaller than a snapdragon 845. The entire point of AMD's plan is they just use those chiplets on everything. But yes the yields on those tiny chiplets is great

The CPU block on a Snapdragon is pretty small though. Keep in mind how much IP there is on those SoCs (GPU, ISP, DSP, modem, NPU etc). It takes up a lot of the space. I do agree though that the chips are getting so small that the size argument becomes almost irrelevant.

Link to comment
Share on other sites

Link to post
Share on other sites

Not low power, 'm talking ultra hgh binned, super highly clocked cores for gaming. Something general ones still can't seem to achieve.

AMD Ryzen 7 5800X | ASUS Strix X570-E | G.Skill 32GB 3733MHz CL16 | PALIT RTX 3080 10GB GamingPro | Samsung 850 Pro 2TB | Seagate Barracuda 8TB | Sound Blaster AE-9 MUSES Edition | Altec Lansing MX5021 Nichicon/MUSES Edition

Link to comment
Share on other sites

Link to post
Share on other sites

AMD DESTROYS Intel with HIGH CORE COUNT and IMPRESSIVE CLOCK SPEED

 

Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler
Spoiler

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

7 hours ago, RejZoR said:

Not low power, 'm talking ultra hgh binned, super highly clocked cores for gaming. Something general ones still can't seem to achieve.

So an FX9590? That sucked at 5Ghz and still sucked at extreme OC 7 Ghz.

 

The problem is Zen/Zen+ on GloFo can't clock higher than ~4.4Ghz, single core or anything. No binning will get you passed that. Games are still rather demanding on the cores they do use so making them weaker in some way to maybe clock higher will net nothing or a loss. Intel's fab processes are just so much better because they are not modified low power mobile nodes like Samsung, GloFo and TSMC are.

Link to comment
Share on other sites

Link to post
Share on other sites

Hang on, the guys complaining about the clock speeds of  AMD chips, isn't it a better idea to get coders to make use of more cores where possible instead to spread the load... that would help keep heat down too, and because the trend is "moar corez" that should in theory sort headaches out for most. I mean you can get an 8 core/16 thread first gen ryzen now for around £150-ish, that if the developers could load balance the games CPU dependency better cope with a fairly hefty game?

Please quote my post, or put @paddy-stone if you want me to respond to you.

Spoiler
  • PCs:- 
  • Main PC build  https://uk.pcpartpicker.com/list/2K6Q7X
  • ASUS x53e  - i7 2670QM / Sony BD writer x8 / Win 10, Elemetary OS, Ubuntu/ Samsung 830 SSD
  • Lenovo G50 - 8Gb RAM - Samsung 860 Evo 250GB SSD - DVD writer
  •  
  • Displays:-
  • Philips 55 OLED 754 model
  • Panasonic 55" 4k TV
  • LG 29" Ultrawide
  • Philips 24" 1080p monitor as backup
  •  
  • Storage/NAS/Servers:-
  • ESXI/test build  https://uk.pcpartpicker.com/list/4wyR9G
  • Main Server https://uk.pcpartpicker.com/list/3Qftyk
  • Backup server - HP Proliant Gen 8 4 bay NAS running FreeNAS ZFS striped 3x3TiB WD reds
  • HP ProLiant G6 Server SE316M1 Twin Hex Core Intel Xeon E5645 2.40GHz 48GB RAM
  •  
  • Gaming/Tablets etc:-
  • Xbox One S 500GB + 2TB HDD
  • PS4
  • Nvidia Shield TV
  • Xiaomi/Pocafone F2 pro 8GB/256GB
  • Xiaomi Redmi Note 4

 

  • Unused Hardware currently :-
  • 4670K MSI mobo 16GB ram
  • i7 6700K  b250 mobo
  • Zotac GTX 1060 6GB Amp! edition
  • Zotac GTX 1050 mini

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

21 minutes ago, paddy-stone said:

isn't it a better idea to get coders to make use of more cores where possible instead to spread the load...

For gaming in particular? Depending on the game type, there may be only so much you can split up the task flow before you start running into data-dependency issues.

Link to comment
Share on other sites

Link to post
Share on other sites

On 11/17/2018 at 9:38 PM, leadeater said:

So an FX9590? That sucked at 5Ghz and still sucked at extreme OC 7 Ghz.

 

The problem is Zen/Zen+ on GloFo can't clock higher than ~4.4Ghz, single core or anything. No binning will get you passed that. Games are still rather demanding on the cores they do use so making them weaker in some way to maybe clock higher will net nothing or a loss. Intel's fab processes are just so much better because they are not modified low power mobile nodes like Samsung, GloFo and TSMC are.

 

 

Someone, (it may have been you), recently pointed out that with the right cooling, EPYC1 and Zen1 products all clock up to about the same level. That suggests to em it's not the cores that are the limiting factor. Potentially with all of that kept on 14nm they may have improved things to the point where it can handle the strain of higher clock speeds. You're not likly to see much benefit on EPYC as the core count upscaling is generally pushing the thermals allready. But the desktop potentially has room.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, CarlBar said:

Someone, (it may have been you), recently pointed out that with the right cooling, EPYC1 and Zen1 products all clock up to about the same level. That suggests to em it's not the cores that are the limiting factor. Potentially with all of that kept on 14nm they may have improved things to the point where it can handle the strain of higher clock speeds. You're not likly to see much benefit on EPYC as the core count upscaling is generally pushing the thermals allready. But the desktop potentially has room.

Don't think it was me, if they had to stay on 14nm they would have to have gotten a custom node from GloFo/Samsung/TSMC optimized for high power/high frequency. Not actually a problem to get that though, Zen based products seem to be selling well enough that it would be something AMD could have gotten done.

Link to comment
Share on other sites

Link to post
Share on other sites

34 minutes ago, leadeater said:

Don't think it was me, if they had to stay on 14nm they would have to have gotten a custom node from GloFo/Samsung/TSMC optimized for high power/high frequency. Not actually a problem to get that though, Zen based products seem to be selling well enough that it would be something AMD could have gotten done.

 

Um, i'm confused. We know the IO die is staying on the existing GloFo 14nm process...

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, CarlBar said:

 

Um, i'm confused. We know the IO die is staying on the existing GloFo 14nm process...

Oh I thought you were meaning about Zen/Zen+ itself or if the cores stayed on 14nm.

Link to comment
Share on other sites

Link to post
Share on other sites

I am missing something completely, the quote reads:

Quote

Now if we compare the chip to the Ryzen Threadripper 2950X part, another 16 core configured chip which has virtually the same core specs as the EPYC 7371, we see much higher clocks at 3.5 GHz base and 4.4 GHz boost. The part isn’t a fair comparison though since AMD is using their slightly optimized Zen+ cores. So if we compare it to the same Zen core based Ryzen Threadripper 1950X, we still get higher clock speeds of 3.4 GHz base and 4 GHz boost.

but this chip's clocks are 3.1 GHz base, 3.6 GHz (16 core) and 3.8 GHz (8 core) boost, mentioning that this is the highest clocked zen CPU yet? Most Zen CPUs clock to at least 3.9GHz even the ones mentioned in the quote have higher clock speeds. What am I missing? 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×