Asus 5700xt Strix Adventures

The plan was to take my powercolor liquid devil to extreme cooling since it was a good performer on water but finding a ln2 pot that fits and due to the 2 caps that's above the core meant that i would have to unsolder and put them on the back of the card. To avoid the unnecessary hassle i put the card for sale and picked up a strix card despite the bad rap with the stock cooler it has the best pcb, so i picked one up from MC testing the card it never overheated so i must've got one from the new batch.

i eventually hit a limit at 2250mhz with only 1.27v on air so i ordered a EK block quick

i installed the block but noticed the thermal pads didn't squeeze out the way they would normally.
also when i put the card in the case the card would separate from the block long story short its a design flaw as the top left corner has no support.




































i decided to test it light load to check temps the core was excellent and the hotspot was on 20 degrees away from core temps but the memory and mem vrm was overheating i placed something to wedge the card up in the corner and the temps did improve. so i shut it down broke the card down and made a case with ek havent heard from the yet but i sent the block back to newegg

decided to just clean the computer out in my down time



after some searching i found that bits power and alpha cool had blocks with support in that corner i picked up the bitspower block and it had pad placement for every heat source on the card plus it allowed me to use the stock backplate..



















got the system together and will test later

 

needed more horsepower to attack HWbot hopefully this 3950x is worth a damn

 

testing clocks vs temps

 

UPD. You RIG

 

testing memory scaling in timespy 1750 1850 1900mhz

 

my test bench finally came

 

my gskill psu failed
picked up a evga




almost finished with the test bench

 

Now thats cool

I have a raijintek-calore-c360d-copper-radiator-360mm and a 120mm rad.

So the loop goes pump - vega 64 - 120mm rad - cpu - 360 rad -pump. Cant keep the temps under 50c for some reason, the ryzen 1600 even goes to 70c at some point well it has 1.4v.

Really dont know the reason . I even have a fan on the vrm lol .

Your 5700xt is amazing. Did you change from your x99 to ryzen?

 

whats your ambient temps like?

And thank you.
my x99 is still in my main rig and the ryzen is on the test bed gotta learn the chip im still a ryzen virgin lol

 

https://hwbot.org/submission/4464189_

https://hwbot.org/submission/4464168_

 

had a good weekend of benching




https://hwbot.org/submission/4467483_

https://hwbot.org/submission/4467493_

https://hwbot.org/submission/4467497_

https://hwbot.org/submission/4464189_

https://hwbot.org/submission/4464168_

 

That gpu when pushed has some scores right on the 2080ti mark. If AMD had released a 300w version with more cus maybe the market was diferent.

Nice scores and grats. And here I am with my crap Vega 64 red devil lol crashes passed 1750mhz.

My temps are better now, I reallized that was some air inside the big rad I cud hear the water droping inside . Now they are a bit better, I think my 1600 die is junk and its not straight, In a few weeks will have a 3900x and then I can play. My x370 carbon pro already has a VRM fan in there waiting for it

Edit: I think that timespy is so Nvidia bios its dx12 and we know that amd is better there, but on the benqmark just favores green. thats the reason I like firestrike is better on that.

 

There's one model with 240 or 260 for the target watts instead of the 185 - 220 that most others hover around going by the TechPowerUp bios database info as I remember, no idea if the actual GPU core can scale up and push that to any significant advantage but I suppose a water block can at least allow some scaling together with the +50% power limit via the software going from a higher bios limit and into the 300 range.

Then there's these spikes and the higher transient draws but a good high-end PSU should be able to cope both with amperage and wattage pushing into near double so it might not be too big of a hurdle just something to keep in mind.
P state modifiers if still doable to go above the default caps and hit above 2200 (2150 effective due to Wattman often being -50Hz from the set value for some reason.) and maybe some further gains although without additional hardware these gains might be a bit more limited than something like a "Big" Navi would have achieved if AMD had made one.


EDIT: Suppose a water block on the Navi20's will be a fun little benchmark plus other improvements both on the hardware and what the newer drivers could have over existing sets.

But that's for some future test or benchmark once those are on the market possibly Autumn or early Winter assuming the various rumors are somewhat accurate.

 

nice and thanks


yeah i got my vega64 in the main rig now trying to remember how to get 1.25v without using the LC bios that causes issues

 

I have tried to use the LC bios and its a no go. Its trootles the card, its driver driven.

With PPT I can set 150% target power, issue is I have to put 1.1mv and 1670mhz max target, the card will boost to 1715mhz like this anything above and its a driver crash.
HBM will do 1200mhz with lvl 1

Still love the Vega I think its a greate card.

 

putting this gpu to work lol


https://hwbot.org/submission/4473595_

 

And thats on water? Good work, did you try and play a game at those setings

If you can make youtube videos with it like that you iwll get loads of views

As a question do you know a fix for an unmolded vega? If there a paste that will fix the diferences.

On your card did you try to change the timings of the vram? AMD Memory Tweak maybe this will help

 

All of this on sub 390€ GPU