Jump to content
Sign in to follow this  
Dasa

AMD Zen

Recommended Posts

AMD promises 32-core Threadripper processor for later this year

 

Mere hours after Intel raised eyebrows with the promise of a 5GHz 28-core processor, AMD has done four cores better with its own tease of a 32-core chip. The second generation of AMD’s Threadripper processors will have a truly extreme variant that straps four 8-core Ryzen dies together to form a unified, humongous 32-core part. This “heavy metal” Threadripper can handle 64 processing threads at a time, doubling the core count and capabilities of the original Threadripper, though it can still fit in the same motherboard socket as the first generation.

 

The only other details that AMD disclosed here at Computex were that the second-gen Threadripper will be built using a 12nm process and will be available to buy in the third quarter of this year.

 

amd_thread_vladsavov.jpg

 


COMPUTEX%202018%20-%20Lisa%20and%20David

Share this post


Link to post
Share on other sites

Wonder how it compares to it's predecessor?

To me it feels like the refresh wasn't gonna be enough bang to cause excitement.

 

I wonder if 32 cores will win it a Mac Pro design from Apple?

Share this post


Link to post
Share on other sites
Posted (edited)

 

Wonder how it compares to it's predecessor?

To me it feels like the refresh wasn't gonna be enough bang to cause excitement.

 

I wonder if 32 cores will win it a Mac Pro design from Apple?

 

 

`Will people fork out 10K for a Mac Pro with 32 core CPU and a Vega GPU?

Edited by Jeruselem

Share this post


Link to post
Share on other sites

`Will people fork out 10K for a Mac Pro with 32 core CPU and a Vega GPU?

You're familiar with Apple customers, yes?

Share this post


Link to post
Share on other sites

 

`Will people fork out 10K for a Mac Pro with 32 core CPU and a Vega GPU?

You're familiar with Apple customers, yes?

 

They'll be idiots who buy those but some won't because ... money

Share this post


Link to post
Share on other sites

They'll be idiots who buy those but some won't because ... money

 

Yeah that's what I said... apple customers

 

:D

Share this post


Link to post
Share on other sites

Ordered 2600x off ebay.. Should be able to sell my current cpu for a bit more than what I paid for the 2600x. Just got it to play with mainly... Might see slight gains in gaming... really just wanted to try something new till next year release and see how it overclocks... Im bad about side grading but if I can get into it without losing much Im game :)

Share this post


Link to post
Share on other sites

Ordered 2600x off ebay.. Should be able to sell my current cpu for a bit more than what I paid for the 2600x. Just got it to play with mainly... Might see slight gains in gaming... really just wanted to try something new till next year release and see how it overclocks... Im bad about side grading but if I can get into it without losing much Im game :)

They reckon that the turbo and xfr are so good that overclocking is a little worse than an all core overclock.

 

If your RAM is samsung B die though... overclock it and tweak the timings to Hardware Unboxed's guide and you should get very noticeable performance gains in games.

Share this post


Link to post
Share on other sites

AMD Zen 2 Based 7nm Rome Server Processors Were Designed To Compete Favorably With Intel Ice Lake-SP Xeon CPUs, Aiming For Multi-Digit Server Market Share Gains

 

AMD has made a grand entrance back in the server market and is going in all guns blazing against Intel’s Xeon processors with their disruptive EPYC platform. While developing their next-gen CPUs for the server market, AMD made assumptions of where Intel’s chips will land in terms of efficiency and performance but Intel seems to be just giving away their market share due to a delayed 10nm process which will put AMD in a huge lead come next generation.

 

AMD has on multiple occasions, confirmed that their 7nm EPYC ‘Rome’ server family is on track. The most recent one was at the one year anniversary of EPYC processors based on the 14nmFF ‘Naples’ architecture. During the webinar, AMD’s Senior Vice President and General Manager of Datacenter and Embedded Solutions, Forrest Norrod, reaffirmed that 7nm EPYC Rome CPUs will be sampling in 2H 2018 and launching in 2019.

 

Forrest Norrod also confirmed future iterations of Zen architecture beyond 2020 which include Zen 4 and Zen 5 while will succeed the 7nm+ EPYC ‘Milan’ processors. However, when designing the EPYC server roadmap, AMD had a few goals in mind.

 

For AMD’s first 7nm server family specifically, AMD made assumptions around Intel’s roadmap and what they would do if they were Intel. There’s no mystery about Intel’s next-generation Xeon CPUs as we know that the Skylake-SP (14nm+) chips will be replaced by the upcoming Cascade Lake-SP (14nm++) family. We have quite a few details regarding the Cascade Lake-SP family which you can check out here but Forrest Norrod has some interesting details regarding Rome.

 

According to him, the AMD 7nm EPYC Rome processors were not designed to compete against the Cascade Lake-SP Xeon family, they were actually designed to compete favorably against Intel’s Ice Lake-SP Xeon processors. You heard it, right folks, AMD’s 2019 CPU family is designed to tackle the Intel 10nm Ice Lake Xeons favorably and things are looking really good for AMD as their Rome CPU family will only be competing against Intel’s 14nm++ server refreshed family, aka Cascade Lake-SP. Intel’s Ice Lake-SP processors based on 10nm process aren’t expected to arrive in the server Xeon space till 2020.

 

“Rome was designed to compete favorably with “Ice Lake” Xeons, but it is not going to be competing against that chip. We are incredibly excited, and it is all coming together at one point.” – Forrest Norrod. via TheNextPlatform

 

There’s no doubt that AMD made a grand comeback in the server space with their highly disruptive EPYC platform. Returning right on time when Intel was at their most fragile position with little to no progress being made towards the 10nm process development, stagnant IPC evolution and very less impressive feature updates on the server side. Sure Purley platform itself was supposed to deliver a good amount of features to consumers but EPYC made that look like child’s play comparison. Just to tell you how much of an impact EPYC made in the server market, Intel’s CEO, said in an interview recently that they are expecting to lose server CPU market share to AMD’s EPYC processors.

 

"Mr. Krzanich was very matter-of-fact in saying that Intel would lose server share to AMD in the second half of the year. This wasn’t new news, but we thought it was interesting that Mr. Krzanich did not draw a firm line in the sand as it relates to AMD’s potential gains in servers; he only indicated that it was Intel’s job to not let AMD capture 15-20% market share.” – Romit Shah, Nomura Instinet

 

 

It looks like AMD is in all-guns-blazing mode, introducing a competitive and powerful lineup gen after gen. AMD is also expecting to witness multi-digit server market share gains in the coming years, up from literally 0% market share before EPYC launched. Unless or until Intel gets things straight with their CPU and core development projects and correct their fab issues, AMD is just going to keep coming back at them with a more powerful response

 

Intel is expected to use a similar MCM approach to AMD’s server lineup with a Cascade Lake-AP part but AMD has the upper hand in both MCM technologies deployed on CPUs to date and they also have the lead in the number of cores deployed. Even if Intel comes out with an MCM die with more cores compared to the current maximum of 28 which will also remain true for the Xeon SP (Cascade Lake) parts, there’s no stopping AMD from deploying a 64 core, 128 thread juggernaut chip on the Rome platform while using current technologies.

 

"Rome was designed to compete favorably with “Ice Lake” Xeons, but it is not going to be competing against that chip. We are incredibly excited, and it is all coming together at one point. We have reintroduced ourselves to the market, gotten the initial traction and wins, we got the initial customer support, and we validated that AMD is a safe choice with an effective processor. With the Rome processor and process, we are going to be in an incredible position going forward.”

 

 

“Our plan for the Naples-Rome-Milan roadmap was based on assumptions around Intel’s roadmap and our estimation of what would we do if we were Intel,” Norrod continues.

 

“We thought deeply about what they are like, what they are not like, what their culture is and what their likely reactions are, and we planned against a very aggressive Intel roadmap, and I really Rome and Milan and what is after them against what we thought Intel could do. And then, we come to find out that they can’t do what we thought they might be able to. And so, we have an incredible opportunity."

 

 

It's good to see AMD taking advantage of a bad situation over at Intel. Hopefully it's the momentum they need to push themselves into a stronger position which will be better for all consumers.

Share this post


Link to post
Share on other sites

Switched out the 1700x for the 2600x to play with. Well, the 2600x is just too good out of the box, there really no point in overclocking, just like what all the reviews say. Sure a few people got up to 4.3+ but its rare. I can't get all cores to 4.2ghz without lots of extra voltage but it boosts over 4.2ghz quite a bit even 4.3 sometimes. Many times I see 2 cores over 4.2ghz at the same time.

 

Cinebench 1700x @ 4ghz:

 

Single: 160

Multi: 1702

 

2600x @ stock:

 

Single: 168

Multi: 1268

 

Timespy 1700x @ 4ghz+?:

 

Best runs were just over 10k, most landed around 9.7k

 

2600x @ stock:

 

9250

 

Firestrike 1700x @ 4ghz:

 

Best just over 21k most in the 20.5k range

Stock around 18.5k

 

2600x stock:

 

20185

 

Temps have not crossed 67c with my loop.

 

The few games Ive tried seem faster when gaming. To me, if being honest, Im frustrated with the fact that I cant really overclock this chip and even the last chip 1700x was limiting, 4ghz was fine at around 1.4 but moving up to 4.1 was not worth it. I still cant run 3200 on ram, with this chip, Ive messed with lots of timing, manual entering and such. The 2600x just seems so good at the new XFR that cant really top it with OC. It's just a bit boring, kinda wish I went Intel but really wanted to go back to AMD as they do have a heck of a product. Ryzen is a great product but Im a tweaker I want to have some room to play and try stuff or even delid if need be. Not that the Ryzen is bad at all though, I would recommend it for sure to most people but for me I want to play and have some fun doing that! I look forward to next years refresh, and will buy it Im sure... I was hoping/wanted to have the week of tweaking the crap out of it at least :) My delid 3770k was so fun to play with. I probably do not make sense to most, but my old barton would OC like crazy lol...

Share this post


Link to post
Share on other sites
Posted (edited)

you tried feeding the ram 1.45v?

if you want to tinker maybe its time to get some b die :) although im not sure how well it will go on a biostar mb...

 

im back playing with my ram again it seems more stable at higher speeds with the latest bios

Edited by Dasa

Share this post


Link to post
Share on other sites

Yea so wish I bought a better kit but back then prices were half what they are now. I wonder if like you said my mb could even go 3200... ram tweaking might be the way to go I'll try some extra volts :)

Share this post


Link to post
Share on other sites

Ok pumping up the ram to 1.45v I can boot and everything seems fine so far, thx Dasa. I had the ram at 1.35v for 2933, really happy to have 3200 :) I guess I should have upped it more sooner

 

I need to research some to see at what point latency vs frequency is best, Im going to look into your charts in you sig again to see what I can see from it too. If possibly or if there is benefit to running higher frequency at the cost of some lost latency then might try it out. Something to actually play with. (I have no clue if it will even boot with anything higher yet)

 

cpu-z

3200 - 16, 18, 22, 53, 75 1T is what xmp profile gives for 3200 which I ran at 2933. 30mins on prime passed.

 

Few other notes, 68c temp is highest with all fans on low and pump at 60% with prime95. GC probably needs to work in some more too. Clocks with p95 range from 3850 to 4000mhz. Jumps a lot with p95 and not all running the same speed. ~1.298v on the core mostly with some jumps, 1.441v the highest but I haven't seen it there but cpuid did.

 

Cine:

SC: 169 (no improvement)

MC: 1318 (decent improvement)

Share this post


Link to post
Share on other sites
Posted (edited)

https://www.overclock.net/forum/10-amd-cpus/1628751-official-amd-ryzen-ddr4-24-7-memory-stability-thread.html

you should find some good info there on what sub timings work well for squeezing more out of ram on amd

 

with amd frequency is king provided timings are not blown totally out the window as the fabric is tied to the memory speed latency reduces as frequency goes up

i use aida64 to keep an eye on the rams final latency

 

turns out my increased stability with the latest bios was just a illusion although it was posting at 1t easier than it use to it still wasn't completely stable at anything over 3866c16

 

a test relevant to actual game performance improvement could be hard

most benchmarks like cinbench see next to no improvement from ram speed

firestrike next to nothing

timespy sees a small improvement maybe 5% when a game would see 20%

maybe the api overhead test?

Edited by Dasa

Share this post


Link to post
Share on other sites
Posted (edited)

Some results with 3200 and aida64 5.97.4600 benchmark. Stress tested cpu, fpu cache and memory for 30mins on aida64. 3.5hrs on p95:

 

3200 @ 16-18-22-53 (as aida64 reports)

read: 47133mb/s

write: 44524

copy: 39661

latency: 71ns

 

3200 seems completely stable!

--

Testing now 3333 @ 16-18-23-55 (as aida64 reports)

read: 48807mb/s

write: 46235mb/s

copy: 40884mb/s

latency: 67.9ns

 

need to run stability tests.

 

Some notes on temps too, they got much higher when running aida64 tests, up to 78c. Cranked up the pump to 75% and fans for rad are at 1100rpm too for these runs. Keeps it around 75c for the highest, and most time not near that, p95 doesnt get near that high even after hours...

 

I tried to post 3466, no go at these timings, it was just to see if it would since 3333 posted :) Ill probably have to loosen the timings but pretty happy so far. Benchmarks show no gains in FS or Timespy.

Edited by gamble

Share this post


Link to post
Share on other sites
Posted (edited)

nice latency is getting to a much better level

im guessing your ram is using hynix chips if so this may be handy

you can use thaiphoon burner to confirm what chips it has

 

https://www.overclock.net/forum/11-amd-motherboards/1624603-rog-crosshair-vi-overclocking-thread-2414.html#post26242714

Few more DRAM timing presets:

Hynix AFR, 1DPC SR

"Safe"

ea5bb6ba_AFR-3200S.png

- ProcODT 60 Ohms
- DRAM Voltage / DRAM Boot Voltage 1.340V (keep these syncronized at all times)
- VDDCR_SOC 1.025V

"Extreme"

24310b0d_AFR-3200E.png

- ProcODT 60 Ohms
- DRAM Voltage / DRAM Boot Voltage 1.405V (keep these syncronized at all times)
- VDDCR_SOC 1.025V

Hynix MFR, 1DPC SR

"Safe"

2ee87dd2_MFR-3200S.png

- ProcODT 60 Ohms
- DRAM Voltage / DRAM Boot Voltage 1.365V (keep these syncronized at all times)
- VDDCR_SOC 1.025V

"Fast"

86e7ba86_MFR-3200F.png

- ProcODT 60 Ohms
- DRAM Voltage / DRAM Boot Voltage 1.390V (keep these syncronized at all times)
- VDDCR_SOC 1.025V

 

or you can find his b die presets here

https://www.overclock.net/forum/11-amd-motherboards/1624603-rog-crosshair-vi-overclocking-thread-2067.html#post26178558

 

these are all for the asus corshair mb though so your mileage may vary

Edited by Dasa

Share this post


Link to post
Share on other sites

Thx Dasa! Yup hynix. I'll give those a try too. I want to see if 3333 and maybe 3466 will be able to stay stable with looser timings too.

Share this post


Link to post
Share on other sites

Thanks once again Dasa :) I used thaiphoon burner and calculator to tweak some settings for 3200. I was unable to get 3333 stable with more volts or higher timings, it would eventually fail some tests or lockup. At 3200 my mb would not let me go any lower than 16, even it I set to 15 it would go to 16. Played with tons of setting and even the safe setting would give some problems. I settled with setting the main 5 timings and working with those, goal was to get to under 70ns and ended with 70.8. All in all it was really enjoyable playing and testing things out. Had a strange thing with reports for my hynix ram as it would give a ? for the AFR or MFR:

 

H5AN8G8N?FR-TFC

 

So, I tried all safe, fast and extreme settings but I guess mine is picky or my mb is as it didnt seem to want me to play with all timing for some reason. Even if I did straight what xmp gave, if I put those in it would lock up.. Strange stuff. I could put in some setting and it would work fine.

 

3200 @ 16-18-18-42 CR1 (as aida64 reports) 2hrs of aida64 stress / p95 1hr

read: 47280mb/s

write: 44891

copy: 40333

latency: 70.8ns

 

Just a side note, Windows was acting kinda weird so I thought I would do fresh install. MB came with m.2 sata3 240gb with 530mb/s read / 410mb/s write. I had bought a mydigitalssd to replace it, but it wouldn't work last year, probably because of how fresh bios and platform was back at release. Thought I would try it again for my os drive and it worked great, super fast, I forgot to test the biostar m200 that came with mb but tested against my vertex4:

 

m.2 pce 3.0 x4, 2600mb/s read / 1300mb/s

 

Test Block Size Result CPU% Drive
Linear Read (Begin) 8 MB 1969.7 MB/s 0 % Disk Drive #3 [bPX] (111.8 GB)
Linear Read (Middle) 8 MB 2644.0 MB/s 3 % Disk Drive #3 [bPX] (111.8 GB)
Linear Read (End) 8 MB 2716.1 MB/s 0 % Disk Drive #3 [bPX] (111.8 GB)
Random Read 8 MB 2237.0 MB/s 3 % Disk Drive #3 [bPX] (111.8 GB)
Buffered Read 1 MB 2448.6 MB/s 0 % Disk Drive #3 [bPX] (111.8 GB)
Average Read Access 0.06 ms 1 % Disk Drive #3 [bPX] (111.8 GB)

 

Linear Read (Begin) 256 KB 425.5 MB/s 1 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)
Linear Read (Middle) 256 KB 433.0 MB/s 0 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)
Linear Read (End) 256 KB 427.5 MB/s 3 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)
Random Read 256 KB 416.3 MB/s 0 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)
Buffered Read 8 MB 357.2 MB/s 1 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)
Average Read Access 0.11 ms 1 % Disk Drive #1 [OCZ-VERTEX4] (476.9 GB)

 

Few other notes for the cpu... I see it boosting all core over 4200mhz quite a bit, it is running a bit cooler as I have upped my fans/pump. It is running over 1.4v+ a lot. Its strange for the manual OC that it doesnt want to play well. Im going to try offsetting again and a few bumps with volts in other areas. I mean if it is running 4250+ I should be able to get all core locked down with maybe 1.45v @ 4.2. When p95 or aida64 is running all cores sit around 4ghz. I see they are going to come out with precision boost OC soon too, might be worth waiting for. Timespy give 9370 with these timings.

Share this post


Link to post
Share on other sites
Posted (edited)

with all those sub timing settings above there may be just a few of them that your mb doesn't like and a few that it does

 

tweaking sub timings make a bigger difference for me than primary

top three for me are

tRFC <mostly this one from ~500 down to 310 but you probably wont get this low with hynix on amd

and i think these two helped a bit but cant be sure anymore

tFAW

tWCL

doubling tREF can help bandwidth a little but its instability is more likely to show up after idle periods than during a stress test so careful with it

 

mind taking a screenshot of what your sub timings are at?

Edited by Dasa

Share this post


Link to post
Share on other sites
Posted (edited)

Sure, please see below:

 

https://photos.app.goo.gl/L4CcD5HXiiimJgn68

https://photos.app.goo.gl/u85uYLKxQ3YMLbFXA

 

 

Still playing around a bit, trying to see if I can lower the timings some and volts to 1.433.

 

I cannot OC all core really at all 4.2ghz will not happen but 4.1 will but with crazy volts, so not worth it. All in all Im still happy with this chip, seems faster for everything compared to my 1700x.

 

I might need to try the trfc like you mentioned, but is at 312 already. I cant believe what good timings you have at that frequency!

Edited by gamble

Share this post


Link to post
Share on other sites

nice you got tRFC a lot tighter than i expected

 

yes b die does handle tight timings nicely apparently it can typically bench around 4000c13 with ~2vdimm

this is what im currently running 1.43vdimm set but its actually reported as being at 1.408v in software

i don't think this is 100% stable though i need to run some overnight tests

 

ram%20speed.jpg

Share this post


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
Sign in to follow this  

×