EK Supremacy EVO Threadripper TR4 Waterblock Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
54,850
EK Supremacy EVO Threadripper TR4 Waterblock Review

EK is one of the most heralded names in the enthusiast computer cooling industry. It has been building waterblocks for custom CPU cooling since 2003. It has now thrown its hat into the AMD Threadripper waterblock ring. Does the Supremacy EVO waterblock have the performance to back up its reputation against tough competition?
 
A rather big disappointment...especially given the better performing and/or cheaper solutions currently available.
 
Looks like they just thought making the braket/block fit was all they needed to do. But didn't consider the fact that the die package is pretty different from a standard CPU. Hopefully they address the issue and make a revision to their block to correctly cool all the dies.
 
That's pretty much what they did. I contacted EK support about the cold plate on the x399 asus monoblock and was told no changes were made. I don't expect much difference in performance but I have it on order and guess I will see.
 
Dekar12 said:
Looks like they just thought making the braket/block fit was all they needed to do. But didn't consider the fact that the die package is pretty different from a standard CPU. Hopefully they address the issue and make a revision to their block to correctly cool all the dies.
Click to expand...


And perhaps consider getting rid of the micro-fin mating surface to improve heat transfer...
 
This article made me really wonder about which waterblock to get for the Vega 64. I am considering the Swiftech atm but can't find enough solid review information on the different ones.
 
It's disheartening to see a manufacturer produce a block claiming it to be the best possible cooling multiple times yet fail [H]ard when it comes to real world testing. EK shot themselves in the foot. What a shame.
 
i do like the EK blocks ive used, but theyre not doing themselves any favors claiming performance that isnt there. kinda winced when i saw the block interior. contact issues aside, the fin array looks, well...unimpressive. id rather them take the time to design a more optimized product
 
Disappointing. However, is it possible that it could be a particular issue of that individual block? After all, the block you were offered came from another user who reported the same issue with temps. Maybe its worth it to test another few samples, check other reviewers/sites/forums who give it stellar ratings are doing differently if anything (assuming they use some sort of real world viable testing), and speak with EK?

I'd hate for there to be any (more) friction between EK and [H] needlessly, if that particular block is defective and/or there is a particular reason why the combo of parts in the loop you used didn't work well together. (ie high pressure vs low pressure, flow rate requirements etc...)? Ideally, this would be the time when EK could send you a block directly and other liquid cooling hardware to showcase it in what they consider "proper" arrangement. If it falls short there well...that's that. Otherwise, it seems strange that EK, while not perfect, has such a solid rep among enthusiasts in many cases (not to mention partnerships with mobo manufacturers to include liquid-centric features/blocks), so for it to fall so short that seems like a major issue I wouldn't expect them to put-up-with. Without any bad-blood, if [H] and others all came to the realization that yes, the Surpremacy EVO TR4 needs a "V2.0" to be able to pull 4ghz w/o throttling, that should be important knowledge. However, if there are any complicating factors EK may simply disregard this, rightly or wrongly.

On a related note, they just recently released a new monoblock for Asus ROG Zenith (and a few other Asus) boards for TR4. I wonder how it will stack up...
 
Last edited:
Very disappointing from a reputable manufacturer! By the way what fans did you use on the radiator?
 
Glad I didn't buy the EK block. My Bykski has been holding up to 24/7 use for over a month now without a single issue. I originally planned to upgrade it later when more blocks came out but now, I don't think I need to.
 
As I have stated I do not like the EK block on my 1950x. It feels slow to wick away the heat. The EK block on my 8700k feels much faster at heat removal but the chip still runs hot overall.

The build quality is nice but EK needs a revision to the cold plate for sure.

Nice review and many thanks Kyle. You confirmed my suspicions.
 
The Bykski looks like the real winner here, odd performance of the sideways vs "correct" orientation, but that does somewhat point to a product that wasn't designed from the ground up for the application.
 
Xaeos said:
Disappointing. However, is it possible that it could be a particular issue of that individual block? After all, the block you were offered came from another user who reported the same issue with temps. Maybe its worth it to test another few samples, check other reviewers/sites/forums who give it stellar ratings are doing differently if anything (assuming they use some sort of real world viable testing), and speak with EK?

I'd hate for there to be any (more) friction between EK and [H] needlessly, if that particular block is defective and/or there is a particular reason why the combo of parts in the loop you used didn't work well together. (ie high pressure vs low pressure, flow rate requirements etc...)? Ideally, this would be the time when EK could send you a block directly and other liquid cooling hardware to showcase it in what they consider "proper" arrangement. If it falls short there well...that's that. Otherwise, it seems strange that EK, while not perfect, has such a solid rep among enthusiasts in many cases (not to mention partnerships with mobo manufacturers to include liquid-centric features/blocks), so for it to fall so short that seems like a major issue I wouldn't expect them to put-up-with. Without any bad-blood, if [H] and others all came to the realization that yes, the Surpremacy EVO TR4 needs a "V2.0" to be able to pull 4ghz w/o throttling, that should be important knowledge. However, if there are any complicating factors EK may simply disregard this, rightly or wrongly.

On a related note, they just recently released a new monoblock for Asus ROG Zenith (and a few other Asus) boards for TR4. I wonder how it will stack up...
Click to expand...


I had 2 EK tr blocks, same temps with either. Infact in Kyle's review he acheived 84 in stock orientation. I was hitting 87 due to my higher ambient. EK bum rushed this product, and according to them they've had the block ready since december of LAST YEAR as they posted on OCN
"We have the SP3(r2) socket water blocks ready for some time now (since end of December). They will be released on time
wink.gif
"

Wouldn't surprised me if they increased the plate size slapped it on a sample running at stock and said okay works great. The block does do a perfectly fine job at STOCK or very minimal oc. It's once you go past 1.3v all core oc the block can no longer handle the heat. I will be posting pics of the monoblock when I get it next week sometime.
 
Xaeos said:
However, is it possible that it could be a particular issue of that individual block? After all, the block you were offered came from another user who reported the same issue with temps. Maybe its worth it to test another few samples, check other reviewers/sites/forums who give it stellar ratings are doing differently if anything (assuming they use some sort of real world viable testing), and speak with EK?
Click to expand...
The block I have was purchased in retail from Performance PCs. I verified authenticity. The block looked unused to me. He showed similar results with the other block. As for system setup, I see no suggestions on the block literature online.
 
Same outfit that had a huge batch of failed plating a few years ago, then lied about it and stonewalled.

I have an example sitting here on my desk.

The bad taste of EK lingers.
 
Which timezone is that EK PR guy from? We need him in here quick! I'm sure this is all just a misunderstanding!
 
Second video was great in showing clearly heat transfer considerations - A+!

Thanks for showing accurately what this block can do compared to what is available - this should help folks decide much better what is better for TR.

The area of the EK micro fins is clearly not fully covering the two dies but I also see the other block machining into the block as superior as well since water will be cooling three sides of the base plate for each slot machined out, while EK the heat has to transfer through one edge (bottom connecting side) of the fin into the block.

The results are incredibly divergent from the XSPC block - I hope EK redesigns this block and use the competition to test it before releasing it.
 
Ahh its good to have more clarification... it does seem there is a potential issue brewing. Nycdarkness, were you and Kyle using similar other loop components when you both found the blocks less than satisfactory even after taking ambients into account etc..? If it is as is you said that essentially EKWB made a block that is great handling stock speeds and/or older, lesser Intel core/heat loads, and just made it physically larger for TR4, it seems that is insufficient. I know that the Supremacy EVO line has been around for awhile (generally recommended), but maybe they really need a redesign in deference to the new generation of higher-core, higher-heat CPUs out there; not to mention the larger chips/heatspreaders on both Intel and now AMD's HEDP? If other users are finding the same thing in real world tests, then hopefully EKWB will start moving on a newer solution, taking user feedback under advisement that this really isn't good enough.

How different was the latest, favorably reviewed XSPC RGB TR4 block in terms of its cooling structures, compared to previous generations, the Intel version of the same generation etc..? I don't recall if that was covered in the review, but it would be interesting to note if XSPC specifically designed the entire cooling apparatus for TR4 or this newer generation of HEDP at least.

Curious about the monoblock when it arrives and any differences in structures/temps from this version. Maybe the monoblock is revised given that it is later announced and brought to market. Looking forward to hearing about it either way!
 
Xaeos said:
I don't recall if that was covered in the review, but it would be interesting to note if XSPC specifically designed the entire cooling apparatus for TR4 or this newer generation of HEDP at least.
Click to expand...
judging by the size of the fin area i'd say they did. the fins alone are bigger than most other cpus. this just looks like ek rushed so they have something to sell.
 
Xaeos said:
Ahh its good to have more clarification... it does seem there is a potential issue brewing. Nycdarkness, were you and Kyle using similar other loop components when you both found the blocks less than satisfactory even after taking ambients into account etc..? If it is as is you said that essentially EKWB made a block that is great handling stock speeds and/or older, lesser Intel core/heat loads, and just made it physically larger for TR4, it seems that is insufficient. I know that the Supremacy EVO line has been around for awhile (generally recommended), but maybe they really need a redesign in deference to the new generation of higher-core, higher-heat CPUs out there; not to mention the larger chips/heatspreaders on both Intel and now AMD's HEDP? If other users are finding the same thing in real world tests, then hopefully EKWB will start moving on a newer solution, taking user feedback under advisement that this really isn't good enough.

How different was the latest, favorably reviewed XSPC RGB TR4 block in terms of its cooling structures, compared to previous generations, the Intel version of the same generation etc..? I don't recall if that was covered in the review, but it would be interesting to note if XSPC specifically designed the entire cooling apparatus for TR4 or this newer generation of HEDP at least.

Curious about the monoblock when it arrives and any differences in structures/temps from this version. Maybe the monoblock is revised given that it is later announced and brought to market. Looking forward to hearing about it!
Click to expand...

Well I have 2 1950x setups.
My ambient at the time was about 26-28c.
#1 1950x on Asus Zenith on a dedciated cpu loop. Swfitech d5 pump, bitspower 250ml res, 360 hl labs gts 3x corsair ml 120s.
#2 1950x on Gigabyte Gaming 7 shared cpu and gpu loop. EK d5 revo pump, EK 200ml Res, EK 480 PE, SP120x4

I first ran both cpus stock while waiting for the EK block, I used 2 h100is in the meant time. With the h100i I could forget about overclocking it wasn't going to happen with good temps. TR has a suggested operating temperature of 68 according to AMD. So keeping as close to that was my goal. When I got my EK blocks I first installed it onto the zenith setup, I noticed right away 60c on stock operation on load and I was quite content. When I started pushing my cpu, I noticed temps were getting into throttling fairly quickly. I aimed for 4.0 and I got it at 1.329 on chip 1 and 1.35 on chip 2. IMO these 2 are good chips as some take 1.4 plus to get 4ghz stable. I first noticed my zenith setup going to 87 in prime and throttling would occur, I thought maybe there were some contact issues. I reseated the block, same thing. Then I thought maybe it's my ac5 which I doubted, so I changed the paste and same thing. I then took the block apart to check and sure enough no problems. At this point I thought I had a bad block and while I had been waiting for a heatkiller block on the second setup, I just went ahead and bought another EK tr evo from PPC. Installed it onto my 2nd setup and sure enough in a open air p5 case same temps. I was hitting 85 instead of 87, and throttling.

I bought the xspc block after reading Hardocp's review where I saw 75c with 1.4v+, I bought 2 xspc blocks and sure enough I dropped 10-12c in prime on both setups.

I have about 5 regular supremacy evos. I've taken them apart for cleaning many times, and it is the same fin layout. I have a XSPC raystorm on my 6700k and I think the evo out does it by a little bit. I had removed the ek evo plexi on that system as the plexi cracked but didn't leak lol. The xspc raystorm neo block for TR is very similar to their other blocks but they enlarged the cold plate as well as the fin area.

Also EK told me via email support,

There was no change to the cold plate on the monoblock, they have no internal testing comparing the mb and tr block either. Their RD dept said performance is good. That was the response pretty much.
 
I am now thinking of selling my EK block and getting the Raystorm instead for my 1950x.

I also have a Raystorm for 115x socket and AM4 but I can't find the 115x bracket anywhere or else I would totally use that block on my 8700k
 
Interesting results but did I miss something? Kyle's video implies that since the microfins cover far less area over the Threadripper, the EK might be less productive. The results imply support that that result might be correct.

The first thing I would do to support that suggestion might result would be the temps of the cores....did I miss them? (I certainly could have missed them.) The cores outside the microfins should be higher if that is true, assuming conduction within the block doesn't resolve the issue. What about a 1900X versus a 1950X and their core locations on the die. Where are the cores on these 2 CPUs and where are they located related to the microfins?

I've spent ~45years working in fluid flow and its associated heat transfer and I have a number of questions as to why the results are what they are. That said, I do NOT question Kyle's abilities nor his results. My question revolves around why what has been a very good water block manufacturer over the years and why the results are what they appear to be.
 
deeppow said:
What about a 1900X versus a 1950X and their core locations on the die. Where are the cores on these 2 CPUs and where are they located related to the microfins?
Click to expand...
All Threadripper processors are the same in terms of die layout.
 
deeppow said:
I realize the core layout on a die would be the same, but which ones are active on a 1900X versus a 1950X?
Click to expand...
I do not know which cores are active on the two live Threadripper dies. But given that the 1920x and 1900x are going to have much less wattage load I am not sure it really matters. That said I think the full footprint blocks are still going to cool better but with a smaller Delta.
 
deeppow said:
Interesting results but did I miss something? Kyle's video implies that since the microfins cover far less area over the Threadripper, the EK might be less productive. The results imply support that that result might be correct.

The first thing I would do to support that suggestion might result would be the temps of the cores....did I miss them? (I certainly could have missed them.) The cores outside the microfins should be higher if that is true, assuming conduction within the block doesn't resolve the issue. What about a 1900X versus a 1950X and their core locations on the die. Where are the cores on these 2 CPUs and where are they located related to the microfins?

I've spent ~45years working in fluid flow and its associated heat transfer and I have a number of questions as to why the results are what they are. That said, I do NOT question Kyle's abilities nor his results. My question revolves around why what has been a very good water block manufacturer over the years and why the results are what they appear to be.
Click to expand...

The core layout in the 1900x doesn’t change. They just disable 1 core from each ccx to get 12 used cores, the 1900x has 2 cores per ccx disabled to get 8 used cores so you still end up with cores outside the micro fin area.
 
This is disappointing. I bought my block thinking it was going to be a top performer, turns out a lot of the difficulties in cooling I've been having could be chalked up to my block.

Too bad EK is the only TR4 block in Australia.
 
Strange - I have the TR4 block and under full load (with a cpu voltage that's really too high) I hover around 60 degrees consistently. I have a loop with a 480 radiator, a 240 double thick radiator and 2 1080tis (not particularly relevant for this). I'm assuming that the values that are being reported are the TDie values which is what the Ryzen Master software reports as well. I must be doing something wrong but curious with regard to what it could be - test below was only running for a bit but have left it running for over an hour with the same result (peaks at 60-61 degrees).

ryzen master.JPG
prime95.JPG
 
Ipseity said:
Strange - I have the TR4 block and under full load (with a cpu voltage that's really too high) I hover around 60 degrees consistently. I have a loop with a 480 radiator, a 240 double thick radiator and 2 1080tis (not particularly relevant for this). I'm assuming that the values that are being reported are the TDie values which is what the Ryzen Master software reports as well. I must be doing something wrong but curious with regard to what it could be - test below was only running for a bit but have left it running for over an hour with the same result (peaks at 60-61 degrees).

View attachment 41517 View attachment 41518
Click to expand...
My guess would be that you are dropping threads. I have tested about 10 different block solutions now and have not found anything that could come close to your temps. Or maybe you have the most awesome TR on the planet. :)
 
Neat. How did you customize yours?

Also, if EK was giving you grief for not providing them with free press how far removed from their favored reviewer list will this place you?

Ipseity said:
Strange - I have the TR4 block and under full load (with a cpu voltage that's really too high) I hover around 60 degrees consistently. I have a loop with a 480 radiator, a 240 double thick radiator and 2 1080tis (not particularly relevant for this). I'm assuming that the values that are being reported are the TDie values which is what the Ryzen Master software reports as well. I must be doing something wrong but curious with regard to what it could be - test below was only running for a bit but have left it running for over an hour with the same result (peaks at 60-61 degrees).
Click to expand...

With what fans at what speeds?
 
Ipseity said:
I don't get it - ran again for 15 minutes or so and checked to make sure no threads were dropping (they're not). Posting the HWiNFO screenshot below - I'm far from an expert on this so happy to post anything else that would be helpful (you can see from the below across the ~15 minutes it was running it peaked at 62.3 degrees).

View attachment 41519
Click to expand...
First your wattage load is 60w below mine. TR 1950x runs up to 360w under full load with the latest version of Prime95 Small FFT. Check to see if your cores are 100% loaded. Also your vCore is over 1.5v. I would consider that dangerous over time.
 
Teenyman45 said:
Neat. How did you customize yours?

Also, if EK was giving you grief for not providing them with free press how far removed from their favored reviewer list will this place you?

With what fans at what speeds?
Click to expand...

Asiahorse Solar Eclipse PWM Mirage 32 LED 120mm - they're running at close to 2000 RPM (they're PWM controlled but let them ramp to maximum cycle which they do with P95 running). I have them set up in push/pull on both radiators so 8 on the 480 and 4 on the 240 (as well as another 4 throughout the case). I've actually been fairly happy with them other than the fact that my computer case is now brighter than most christmas trees.

20171030_210801.jpg
 
Did you mess with sensmi and which revsion of bios are you on. There was a bios on the zenith that reported temps way lower than actual. It was one of the betas. Your temps of 60 under load @ 4ghz is just not plausible under regular watercooling unless you have the most golden TR ever.
 
nycdarkness said:
Did you mess with sensmi and which revsion of bios are you on. There was a bios on the zenith that reported temps way lower than actual. It was one of the betas. Your temps of 60 under load @ 4ghz is just not plausible under regular watercooling unless you have the most golden TR ever.
Click to expand...

Never messed with Sensemi, using UEFI 0801. I'm pretty sure I didn't get the most golden TR ever since it's not fully stable at 4.0 at anything less than a 1.45 vcore.

Happy to test/post anything that would be helpful - at this point I'm legitimately curious.
 
Ipseity said:
Never messed with SenseUi, using UEFI 0801. I'm pretty sure I didn't get the most golden TR ever since it's not fully stable at 4.0 at anything less than a 1.45 vcore.

Happy to test/post anything that would be helpful - at this point I'm legitimately curious.
Click to expand...

Your temp reading is off. I have the same board as you and I am running 1.35 max on load on the 1950x to get 4 ghz stable. I have a 480 hwlabs gts with a d5 for the cpu loop and 4 ml120s on it (one of the best fans out there atm in terms of performance). It's odd that your showing such a low temp something is definitely not right. I would try flashing the latest beta then doing a cmos clear and re-entering your settings.
 
  • Like
Reactions: 10e
like this
You must log in or register to reply here.
Back
Top