cancel
Showing results for 
Search instead for 
Did you mean: 

Rig Showcase Discussions

kxuping
Forerunner

Re: Oscar Mike

Even if they are reference? They are the same AFAIK (stickers differ).

My order was "out of stock" cancelled, because I had to enter cc numbers and they sold out at that moment.

Even got an email saying that my purchase will be cancelled. That night I received an email saying that my order was shipped and my cc charged. So I got one . It was just delivered. Now one week to receive it on my country.

colesdav
MVP

Re: Oscar Mike

What a beautiful machine and an impressive piece of engineering.

Could I please ask you if you have a copy of 3DMark Firestrike?
Given you are running with a Ryzen 7 1800x you should be able to get a very good Physics Score.
Normal RX480's (Air Cooled)  in Crossfire (DX11) or MultiGPU (DX12) should be able to achieve a high graphics score.
Your RX480's are watercooled, so that should help achieve an even higher score.

Is it possible you could try to post  3dMark Firestrike (DX11) and Time Spy (DX12)  results for this machine please?
Thanks

whiskey-foxtrot
Forerunner

Re: Oscar Mike

Thank you colesdav​.

The RX 480's cooled, does allow me to push the core to the limit without going above 36C, but if you compare my scores you'll notice a trend on DX12: The higher the core, the lower the GPU score: Result

On DX11 (Firestrike) it's actually the opposite: GPU scores stay close around the same whether GPU core is overclocked or not, while CPU scores take a bit of a hit. Here's a comparison of 4 scores: Result

colesdav
MVP

Re: Oscar Mike

Hi,

Thanks!

It is very early in the morning here so I might have got this wrong, but it looks like your GPU Memory Clock dropped on your third TimeSpy (DX12) Result with the low score. Was this on purpose? Do the waterblocks cool the Ram Correctly?

Score                          7625           7811           5433

Core clock                 1,450 MHz 1,450 MHz  1,455 MHz

Memory bus clock      2,250 MHz 2,250 MHz  2,000 MHz

0 Kudos
whiskey-foxtrot
Forerunner

Re: Oscar Mike

That was most likely after a reset and I forgot to bump up the Mem clock. I randomly grabbed that score as well since I have a test with just about any variation stored on 3DMark, especially when a new driver comes out.

The mem clock has the biggest impact on the GPU scores (both in games, and benchmarks) as you can see. It definitely has a bigger impact than bumping up the GPU clock and in most cases a GPU clock boost can have a negative impact on scores for both benchmarks and games, especially when DX12. This is just an observation and looking at the raw numbers.

Here's the Result again - the second to the last score is one I ran today with 2250MHz, and the last one in the comparison is the one with mem clock dropped to 2000MHz. As for the cooling, these blocks have full contact with all the critical parts. Looking at it with a Flir gun (infrared camera) does not show any hotspots at all.

fzfanboy
Adept II

Re: Oscar Mike

Wow! That's some serious scores!

darkwolfca
Miniboss

Re: Oscar Mike

congrats man and love the benchmarks.

whiskey-foxtrot
Forerunner

Re: Oscar Mike

Thank you darkwolfca​ I've been holding back and waiting for Vega to finish up the mod. This will have to wait as I'll be gone most of next week for QuakeCon. Finally get to meet some awesome folks from the community and AMD so I'm stoked about that!

whiskey-foxtrot
Forerunner

Re: Oscar Mike

Yes this was a frustrating sale period too. My local store had enough in, but on opening day they only listed 3; now they're releasing them a couple per day. Oh well. On a side note, EKWB did send me a "little package"

20170821_175004-s.jpg

These boxes are HEAVY, and it isn't the packaging material either. I don't think I'll be able to run these in Crossfire, but I'm still going to test it and use the 2nd Vega for my wife's system.

And as far as the reference boards being the same: I'm not too sure on that. Technically they are the same and maybe they all use the exact same board. I just don't want to risk it based on different kind of molding etc. For instance Samsung vs Hynix memory, and resin vs epoxy molding. Apparently there's a  comparison between Vega cards (the GPU's specifically) which have Samsun and some have Hynix. Source is "Videocardz".

whiskey-foxtrot
Forerunner

Re: Oscar Mike

I ran a quick test on another PC (since with Oscar-Mike you can't do a quick swap due to the water-cooling). I ran one of my RX Vega 64's out of the box against a GTX 1080. I can't wait to have the cooling blocks on this so I can really play with fine-tuning!

I also included a comparison with my Oscar-Mike system (RX 480's) which still trades blows with both the RX Vega and GTX 1080: Result

While Radeon RX Vega currently may not look as favorable (in raw numbers, not actual performance) compared to other brands, keep in mind that only a small percentage of the RX Vega is currently put to use. Think of the AMD's Fine-Wine Technology - it will get better over time as more applications adapt to the power of Vega! I would highly recommend any RX Vega if you can afford it, without any hesitation.