-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
The Radeon VII Owners Thread
- Thread starter Kaapstad
- Start date
I feel kinda like an idiot as I did have a decent size bubble there. It formed on the outer side of the block towards the PCIE power connectors. Was quite difficult to remove too but I suspect this is the case because there is very poor flow going down that path. In any case temperatures did improve by about 5C which is still not great. Junction hits 105C at stock from Unigine Superposition still or 80C if I'm using my 960mV profile.The 3mm backplate pads will loosen, as the manual says the screws need checking after 12 hours.
Have you got all the air out of the block too, my card to vertical so wasn't too bad.
I'll have to give the block another look tomorrow to see if I can tighten those screws but I'll need to purchase an angled screwdriver as my tubing blocks access currently.
Multiple reasons. It is likely to age better as you mentioned since it has 16GB of VRAM. Other reasons are that I disagree with Nvidia's recent practices namely GPP and pricing/price creep every generation. Whilst the Radeon VII is also just about as expensive the price can be rationalised by the component cost. My final reason is the ability to tweak the card. Radeon cards still allow you to tweak the voltage, tend to have greater potential, and also have increasing overclocking support incoming such as the HBM timing tuning that is soon to arrive. Nvidia is trending in the opposite direction, hence GPU boost and limiting voltage on Pascal cards to what they run at stock (1063mV). Then there are powerplay mods of course which don't require you to change the bios (big plus).
Edit: Forgot about segmentation of A and non-A tier chips ie them shipping out binned GPUs as their FE cards to get good reviews and then charging extra to board partners for those so those custom ones cost even more ridiculous amounts.
Last edited:
Permabanned
I took a few things into account:
1, I do not like how Nvidia works or how they treat consumers.
2, At the time I was going to buy a new gpu Nvidia stated they were allowing their gpu's to use adaptive sync through Freesync monitors, that should have worked in Nvidia's favour but they turned around & said that out of several hundred Freesync monitors tested only 5 worked properly with their gpu's, I think that was a lie to save face over contradicting their original superior tech claim but my monitor wasn't one of the 5 so I wasn't going to chance it, I know my monitor works perfectly with AMD gpu's so why should I?
3, Performance wise the 2080 matches & directly replaces the 1080ti but it has only 8gb's of memory compared to the 11gb's on the 1080ti. Regardless of it being GDDR6 compared to the 5x on the ti it didn't sit right with me and the Radeon VII coming with 16gb's of HBM 2 looked a lot more appetizing.
4, AMD have recently started supporting VSR for ultrawide monitors & I like being able to push my resolution from 3440x1440 to 5120x2160 in the games I have the grunt to do so.
5, I spent £650 on a Sapphire VII, at that time it would only get me one of the cheapest 2080's & from what we're hearing now that would have meant I'd bought one of the inferior B models if I had so in hindsight that's another negative for Nvidia and another reason why the VII was the right choice for me.
That's why I went with the VII & I've had no reason to regret it so far.
Associate
I did some reading and the VII seemed to fit what I wanted at the time. I don't think we are going to see another card like it for quite some time.
nVidia were having problems with GDDR6 (which seems more likely to be a controller issue) on the 2080 and 2080Ti so there seemed to be less of a gamble going with the Radeon and it's HBM2.
Personally every time I upgrade, and it's not often, I look for at least double the performance of my existing card. My previous card was the KFA2 970GTX, prior to that the 670GTX, prior to that the ATI4890. I like to get my moneys worth.
Associate
From your pictures I'd say your not getting good contact in the middle of the die. I've used HE Extreme on mine and it came with a spreader. I spread a good layer evenly over my die then applied the block. Removed the block before using screws to check contact and had nice contact across the whole die.
The backplate will get warm as the back is where the hottest part of the die is.
I did some reading and the VII seemed to fit what I wanted at the time. I don't think we are going to see another card like it for quite some time.
nVidia were having problems with GDDR6 (which seems more likely to be a controller issue) on the 2080 and 2080Ti so there seemed to be less of a gamble going with the Radeon and it's HBM2.
Personally every time I upgrade, and it's not often, I look for at least double the performance of my existing card. My previous card was the KFA2 970GTX, prior to that the 670GTX, prior to that the ATI4890. I like to get my moneys worth.
You all make some very good points. I'm selling my 1070 and trying to decide which way to go. Nvidia are just way too expensive for me atm. I don't fancy paying £500 for a second hand 1080ti do it might be an RVII to tide me over until next gen Radeon arrives. Is there much difference in the brands? I usually go Sapphire for AMD GPUs.
Right now they're all the same as we only have reference cards. I've seen hints at a Powercolor Red Devil card but not sure if it will come out.
Alright. I'll have to give it another look if tightening the block doesn't help. I honestly thought the paste was that way because the block was convex or something, hence pushing the paste out onto the edges (I did minimal spreading). How much paste did you use anyways? As you can tell I used most of the tube =P
Associate
I did use the whole tube, just scrape off the block onto the die to reuse it. It needs to look like this.
Associate
Just a pair of fresh figures on the alphacool block performance. Worst case using FS ultra stress test!
New HWinfo out too beta 3740.
Stock clocks with 948mv:
Sweet spot setting, 1901mhz, 1150mhz HBM 1000mv:
New HWinfo out too beta 3740.
Stock clocks with 948mv:
Sweet spot setting, 1901mhz, 1150mhz HBM 1000mv:
Associate
I did use the whole tube, just scrape off the block onto the die to reuse it. It needs to look like this.
Yep, that's the method I used too.
Associate
It's a lot of paste, but necessary on this package. What temps are you managing with your Alphacool block?
Associate
I hit a peak of 83 running Superposition with an average of 52. While gaming a lot less though, generally in the 50's.
Soldato
Why so much paste? Doesn't that indicate you are not getting good contact?
From the GN video I saw the surface looked uneven, think the HBM2 sticks up a bit.
Associate
You've basically got five dies in one, the core and four HBM2. They are not generally even so more paste is needed. Also that is the stated method in the instructions.
Soldato
Fair enough. I take it that paste is non-conductive?
Associate
Playing around with the OCs, my GPU isn't the best as the stock voltage is 1101.
2000mhz doesn't seem possible for me. 2000 @ 1150 crashes and I don't really want to push the voltage any higher than that.
Weirdly, I can run 1975mhz/1090mv + 1150mhz HBM and it's fine. Strange that I can't get an extra 25mhz with 60mv but that's good enough for me.
GPU: 44c, Junction: 66c
2000mhz doesn't seem possible for me. 2000 @ 1150 crashes and I don't really want to push the voltage any higher than that.
Weirdly, I can run 1975mhz/1090mv + 1150mhz HBM and it's fine. Strange that I can't get an extra 25mhz with 60mv but that's good enough for me.
GPU: 44c, Junction: 66c