Masada02

Scoreboards updated!

Boards are updated.

6 minutes ago, Seizan said:

tryed to fire up an old 64 i had lying around, well i was pretty sure is was an 64...

but result on superposition it is named as a 56?! so hmm is me or superposition that is lost? lol

(the 4440 is a non OC, just did a test in balanced mode)

Ya it does that. For whatever reason it can't tell the 56 and 64 apart. Same with the 580/570 and 480/470. That's one of the reasons GPUz is required.

One of these days I'll update the scoreboard... travel hasn't been helping.

2 hours ago, themctipers said:

mine was done at i think XMP memory 

my 3600 seems to be a shitty performer, but oh well, it's still 2-3x faster in games than my 3770k

why the fuck is 1080p extreme harder to run than 4k? 

Because Extreme shaders. 4K optimized uses a different shader profile that is a lot easier to run.

4 hours ago, themctipers said:

custom powermod (+100% power limit, +>300MHz HBM OC)

1337GB/s memory bandwidth btw

Dang, that might be slightly faster than my VII...

Is that with a water block?

Oh look, he finally updated the scoreboards!

12 hours ago, Dacien said:

 

I see GPUz isn't reading the clocks yet for the 5700. Any idea what core clock you were running at?

updated!

@clb92 I need a screenshot of GPUz with your score so I can put you on the board  

updated again  

Boards updated.

Some new old stock Model F boards...

Time to outfit the F122s with brand new springs and flippers and keycaps. 

1 minute ago, Jon Jon said:

I guess I didn't phrase myself well.

 

Ignoring the obvious, it's always something I think about and it's something that we all do when we shift between generations.

 

Would it be worth it to invest into Vega for that performance, or into the new architecture for more optimizations and longevity.

 

It makes me think of people going out and buying GTX 980 Tis when the Pascal cards were out, just because the 1080 was hard to find and it was still faster than the 1070.

 

I guess that's where I want to steer this conversation.

 

Are the new RDNA optimizations worth it while being slower today, or would a Radeon VII be just better strictly because it is overall faster in today's games?

 

Heck, the same conversation can be had with Pascal versus Turing. Is it even still worth it to go Pascal, or just go with the Ray Tracing or non Ray Tracing Turing, even if the Pascal card may today yield better performance numbers depending on the model?

It's a good question. Generally when I hear "optimizations" I think they mean they cut out unnecessary features. I'd bet that the RDNA architecture has a lot less of the compute/workstation features that makes Vega so versatile. Though with RDNA you get the strong IPC uplifts I guess. AMD did say at one of their last shows that they would still be using the Vega architecture in workstation cards going forward, and that RDNA would be focused on the gaming market. I take that as the support for Vega won't stop for a while.

Hopefully they will enable some of the Vega features that we haven't seen yet (like rapid packed math) whenever the RDNA cards start to show up since they support them as well.

I'm curious how well the 5700 chips will OC too. Tweaking on the VII is fun already and it looks like the regular 5700 will be a strong overclocker. Could be fun just on it's own.

20 minutes ago, Dat Guy said:

Damn! Where do you find those in that condition?

To be fair, I cleaned and restored both so they weren't that nice looking when I originally got them. Generally speaking, unless things are physically broken, these clean up really well.

This is how dirty the barrel plate was originally...

And another F122... the deal was just too good to pass up.

35 minutes ago, X-System said:

I checked if Unigine 2 Superposition is multithread or not. I see this benchmark isn't 100% multithread

 

I would like this benchmark to be multithread like Assassin's Creed Unity who is 100% multithread. The score would be better, I think.

It basically doesn't use any cpu power so multithreading isn't super necessary. I think there's a slight advantage to higher cpu clocks, but realistically it doesn't care. It's all about the gpu here. 

Been doing some tinkering on my F122. Switched a pristine set of one-piece caps on from my F AT as well as some stems for the non-alpha keys from a donor Model M I had lying around (to fix some scratchy keys). The one-piece caps made a huge difference in the sound, more so than I had expected.

Well I broke down and bought a Radeon VII gold edition.

Not really planning to do any serious OC attempts right now, so for now here's just some basic tweaking in Wattman.

20 hours ago, Madgemade said:

So, I recently saw a Titan Black Hydro Copper on eBay for cheap. Apparently it crashed under load so I got it for $80 shipped.So much for crashing under load. This old GPU sure runs cool compared to the Radeon VII. Never went above 32C!

CPU used is the Ryzen R3 1200 (clocked at 3.8Ghz), but with RAM as old as the GPU and running at only 2133.

I did try pushing for 1400Mhz but it wouldn't go past 1370Mhz no matter what, even with custom BIOS mods and no TDP limit, didn't get much over 300W anyway, lightweight by modern standards!

I have a block for the VI now so might be able to bump my score for that up at least a little bit...

That's a pretty sweet deal on that card.

1 hour ago, ch3w2oy said:

It wasn't easy lol. Once I saw how close to the CF score I was, I couldn't stop.. Its addicting. Just downloaded 3dMark so I'm going to see how Firestrike and Timespy perform next.. 

Also, I'm pretty sure 2362mhz is the fastest core clock we've had on superposition.

5 hours ago, ch3w2oy said:

OOPPSS.. Just kidding.. Ignore my post above.. Best Radeon VII and best overall AMD score!

 

I'll get 6700+ another day, I'm tired..

 

Damn! I'm glad someone finally beat that old Fury crossfire score. Too bad crossfire isn't supported on the vii...

I'd imagine yours is one of, if not, the best vii scores in the world. The unigine leaderboard is only for people that purchased a license so it's not comprehensive.

3 minutes ago, Seizan said:

Wait what someone beat my vega 64?!;;

Just barely, lol

1 hour ago, blk_rbbt said:

Yep, I've already done that and at driver 417.22 I was able to up the power to 85% and not have issues. What I don't understand is how in 3d mark it doesn't crash regardless of voltages, power, core, or memory clock speeds. I know I'm not getting any thermal issues. I'v never seen the card go gover 59c even with it OC'd. I'm going to play around tonight with voltages and see if I can narrow down anything else. I hate to rma the card mainly because I'm lazy and I don't want to remove the blocks and install the stock cooler back on and be without a card BUT I also don't want a dud. 

Ya that's weird. It sounds like some sort of software issue, but trying to pinpoint that is so difficult. I've had similar problems with multiple nvidia cards and it eventually gets fixed with some manner of updates and reinstalling of programs... generally after I'm already super pissed off. Having said that, even if it's a software problem and it can't be solved through the normal methods I'd still return it. You shouldn't have to go through hell to make a brand new product work correctly.

7 minutes ago, blk_rbbt said:

yes and I have no idea why. It seems from the GPU-Z logs it happens if voltage goes over 1v. I can set voltage to cap out at .975mv and it's fine regardless of power. This only happens in games and any unigine . For some reason in 3dmark I haven't had one crash other then when I OC'd legitmately too much. 

 

I am using a pci riser cable. Thermaltakes 300mm. If I wasn't water cooled I would slap that card directly into the slot to see if that makes any difference and I suppose I should try that just to rule it out. 

 

I'm just wondering if I'm having legit power issues or the card is junk. 

 

 

The first step I would take is to use DDU and fully clean your drivers off, then reinstall them. I know it sounds dumb and like it shouldn't relate, but I've had my Nvidia drivers cause similar issues in the past. 

If that doesn't fix it, I'd be returning the card. The 2080ti already has a reputation of being unreliable (I had to return mine as well) and the likelihood of your PSU having an issue supplying power seems pretty low.

2 hours ago, blk_rbbt said:

Yep. I've tried just about everything including different cables. 1 vs 2, old drivers (which actually were more stable.) Only thing that seems to work is lowering the power limit. I'm going to set a voltage curve that maxes just under 1v to see what happens. I don't know if it's my PSU or card. 

Wait so your card crashes when it's at the stock power limit?