EK Confirms 4.2GHz easily attainable on Threadripper

[done12many2]

Just now, AnonymousGuy said:

Could someone TLDR: what does "socket got hot" mean?  Backplate?

 

Then entire socket to include the backplate was heating up when he overclocked.  He was able to take care of it with 2 x 120mm fans.  No big deal.  

[Jahramika]

2 hours ago, Damascus said:

4.2 is probably XFR folks

XFR is not what EK considered Overclocking

[AnonymousGuy]

Just now, done12many2 said:

 

Then entire socket to include the backplate was heating up when he overclocked.  He was able to take care of it with 2 x 120mm fans.  No big deal.  

Bam, already got 'dis :

 

[done12many2]

Just now, AnonymousGuy said:

Bam, already got 'dis :

 

 

You sir are adequately prepped for TR overclocking.  

[Damascus]

1 minute ago, Jahramika said:

XFR is not what EK considered Overclocking

How do you know?  Look at the reviews, XFR ie basically the only way to get past 4.0ghz without pushing stupidly high voltages 

[AnonymousGuy]

4 minutes ago, done12many2 said:

 

You sir are adequately prepped for TR overclocking.  

Wonder if this is because -X has FIVR and TR doesn't.  TR has to deliver all of the power at a lower voltage higher current thru traces on teh board whereas -X can drop lower current higher voltage thru the board to the package.

[goodtofufriday]

@leadeater @tom_w141

 

Maybe this is how.

 

[MoonSpot]

2 hours ago, tom_w141 said:

Yeah nearly 2 full hours wtf lol. I mean I'm at work so I don't to do it as I've already got 1 thread in the news section today generating plenty of notifications  I need less distractions not more

All the video reviews I've seen thus far are incomplete. All the reviewers seem to be splitting up the review coverage into multiple videos, and surprisingly they seem to all be covering the same points.  So I've stopped streaming them.  Going to have to look for written articles and square away time to dive in.  Was expecting to listen to reviews in the background, but they seem to be particularly shallow this time around; plus we've had quite a bit trickle in over the preceding X weeks.

[Bananasplit_00]

2 hours ago, The Benjamins said:

Ya the extra 200+W for going from stock to 4.0

Holy shit that's crazy. 

[goodtofufriday]

15 minutes ago, AnonymousGuy said:

Bam, already got 'dis :

 

thats.... thats brilliant

[done12many2]

9 minutes ago, AnonymousGuy said:

Wonder if this is because -X has FIVR and TR doesn't.  TR has to deliver all of the power at a lower voltage higher current thru traces on teh board whereas -X can drop lower current higher voltage thru the board to the package.

 

That would make complete sense. 

 

Unless there are some changes made, I don't thing we'll be seeing any higher core counts on this socket.  Especially any higher core counts running at 4 GHz.

[AnonymousGuy]

1 minute ago, goodtofufriday said:

thats.... thats brilliant

Ended up not mattering (this was with a 7700K) at all.  I saw more of a difference switching from CLU to Conductonaut

 

Oh and you have to completely mod the case to fit a waterblock + hose behind a motherboard

[done12many2]

Just now, goodtofufriday said:

thats.... thats brilliant

 

And more space efficient than trying to figure out how to get a 120mm fans to call the backplate for the TR overclock. 

[goodtofufriday]

2 minutes ago, AnonymousGuy said:

Ended up not mattering (this was with a 7700K) at all.  I saw more of a difference switching from CLU to Conductonaut

 

Oh and you have to completely mod the case to fit a waterblock + hose behind a motherboard

Might fit better with something like this old school joint

 

[Bananasplit_00]

2 hours ago, The Benjamins said:

Hybrid liquid cooling, do a standard loop and then submerge it in mineral oil.

Oh man that's another thing I want to do... Sounds amazingly fun. 

 

2 hours ago, The Benjamins said:

Also due to it being in oil you can run the water loop through a chiller with out any worries of condensation.

 

I think I just solved x299,x399 OC heat problems.

... You did... You could run a phase change without worry! 

 

1 hour ago, Swatson said:

4GHZ 32CORE CHIPS LETS GO GO

800W HERE WE COME!!!! 

 

27 minutes ago, AnonymousGuy said:

Could someone TLDR: what does "socket got hot" mean?  Backplate?

The metal it sits in, the part on the board. 

[The Benjamins]

Just now, Bananasplit_00 said:

Oh man that's another thing I want to do... Sounds amazingly fun. 

 

... You did... You could run a phase change without worry! 

 

800W HERE WE COME!!!! 

 

The metal it sits in, the part on the board. 

Patent Pending™

[Bananasplit_00]

25 minutes ago, AnonymousGuy said:

Bam, already got 'dis :

 

 

Hmm how much does that help? Might pick up a cheap block used and do that on Northern Bee if I can if it helps noticeable amounts. 

[AnonymousGuy]

3 minutes ago, Bananasplit_00 said:

Hmm how much does that help? Might pick up a cheap block used and do that on Northern Bee if I can if it helps noticeable amounts. 

None on 7700K, TR who knows it may help.  I had to put thermal adhesive tape under the backplate, and then thermal pads between the backplate and the waterblock.   My first post here has pictures of how I built it up: 

You would definitely need a side-exit barb if you don't want to leave the back panel of your case off.

 

 

[Sauron]

3 hours ago, Ravager911 said:

According to AMD all Threadripper dies are binned Zen-dies.

That's not hard to believe considering they can choose which go in ryzen 3-7 and which go into threadripper however they want. 8 core dies that are not as well binned can just as easily be used for r7

[The Benjamins]

1 minute ago, Sauron said:

That's not hard to believe considering they can choose which go in ryzen 3-7 and which go into threadripper however they want. 8 core dies that are not as well binned can just as easily be used for r7

It was stated that Threadripper get the top 5% from the fabs.

[AnonymousGuy]

1 minute ago, The Benjamins said:

It was stated that Threadripper get the top 5% from the fabs.

Problem is that's not too helpful of a figure.  Silicon has speed and it has leakage and they follow an exponential curve distribution. We don't know what blend of leakage to silicon speed they're targeting to say "top 5%".  Could just be top 5% for lowest leakage to  fit in TDP for all we know.

[The Benjamins]

Just now, AnonymousGuy said:

Problem is that's not too helpful of a figure.  Silicon has speed and it has leakage and they follow an exponential curve distribution. We don't know what blend of leakage to silicon speed they're targeting to say "top 5%".  Could just be top 5% for lowest leakage to  fit in TDP for all we know.

true, but it is %5 of the silicon meets the requirements for TR4, the rest would be for AM4.

[done12many2]

1 minute ago, AnonymousGuy said:

Problem is that's not too helpful of a figure.  Silicon has speed and it has leakage and they follow an exponential curve distribution. We don't know what blend of leakage to silicon speed they're targeting to say "top 5%".  Could just be top 5% for lowest leakage to  fit in TDP for all we know.

 

Yeah, binning for either only leads to extremes in one thing or the other.  You either end up with a slower chip regardless of the voltage you throw at it or a fast chip that just loves lots of current.  I would imagine that binning for speed won't really be advantageous for TR, so they are probably favoring leakage.

[The Benjamins]

@AnonymousGuy @done12many2

 

I found the quote

Quote

The Threadripper 1950X offers a 3.4 GHz base frequency that boosts to 3.6 GHz during heavily-threaded workloads. Mainstream Ryzen models feature a dual-core Precision Boost during lightly-threaded tasks, but due to Threadripper's dual-die design (which we'll cover on the following page), that expands to a quad-core 4.2 GHz boost. Much like Intel's Skylake-X series, AMD offers unlocked multipliers for all Threadripper models, but improves overclocking frequency and voltage scaling by selecting the top 5% of Zeppelin dies. That should equate to lower voltage requirements compared to Ryzen 7 models.

http://www.tomshardware.com/reviews/amd-ryzen-threadripper-1950x-cpu,5167.html

[AnonymousGuy]

Just now, The Benjamins said:

@AnonymousGuy @done12many2

 

I found the quote

http://www.tomshardware.com/reviews/amd-ryzen-threadripper-1950x-cpu,5167.html

Thanks.  Sounds like they're shooting for low leakage parts.  5% is a fairly wide net....the best 7700K's were parts that were in the top sub-1% of parts.

 

Don't even get me started on bullshit numbers like ASIC Value   Most meaningless number ever.