Originally Posted by ebduncan
well my r9-290 ran at 1165mhz core/1500mhz mem stock voltage under water. At those speeds it was scoring well over 14k gpu graphics score in firestrike p, basically in line with stock 980's and fury (non x).
I had a xfx rx480 reference 480 for a bit for a review, and it was slower than the r9-290. Not by much but was slower overall. Granted the 290 had a waterblock on it, and the rx-480 was reference blower.
I wanted to avoid having to buy a mid-grade card to keep me going until something that can push 4k well which is currently the titan xp, and I don't exactly want to pay that price. Hey nvidia this is your que to send me a review copy :-D. I just wanted to ride my hawaii heat wave into the future of Vega and Volta before dropping cash on a replacement card.
I see what you're saying...I was on that Hawaii train for quite some time as well. I sold 290 out of my HTPC when the 480's were coming out just so I would have something new to play with..and then grabbed XSPC Blade block since my HTPC was water ready for the GPU. If you're looking for cost effective solution to bring out the true performance of the 480 then the XSPC block does a great job keeping things cool/quiet. If you're going to flash BIOS and try crazy things with voltage/power limits, then the Blade block is definitely inferior for VRM cooling...but it is serving it's purpose for me within the constraints of stock BIOS limitations. I'm sure I could push it harder if I wanted, but this PC is driving my living room TV which is mainly for movies and my wife to play games occasionally. I slowly upgraded it over the years as an excuse to have something else to tinker with, which ended up essentially matching my main rig in terms of specs. Here's Firestrike for comparison:
Sapphire reference w/ XSPC Blade Block & Fujipoly Ultra VRM pads
1400-2250 +24mV +50% Power Target
4790k @ 4.6GHz | 16GB DDR3 2133MHz
Now if I turn Tesselation OFF and compare to my best FS score to date with my 290 here's how it looks: http://www.3dmark.com/compare/fs/10250608/fs/8023532#
290 leads by 5.3% when comparing GS. RX 480 GS: 15,772 ~VS~ R9 290 GS: 16,605
This is the 290 running at 1285/1749 with a modded BIOS that altered the VRAM timings; it's running 1250MHz timings even at 1749MHz . Those are clocks that are typically out of reach for the average Hawaii card, plus a BIOS edit.
I would say that while the 480 is still losing, the performance it is capable of is impressive for the price. Buy an AIB card where they're already running in the low 1300 core range, increase power target, alter voltage a little, and very similar performance to what I achieve is definitely within reach.
I also just swapped out my awesome 290 that was in my main rig for a Fury X. The sale price of $380 was quite tempting, and then I saw that EK & PPCs had the waterblocks on close out. I paid $66.36 for plexi/nickel block, totaling under $450 for card and block. I'll have to run some more benches to see how it fairs against my 290, but from looking previously it should be big enough performance boost to feel.
Here's something I calculated back in February of this year when I was contemplating a Fury: http://www.overclock.net/t/1436497/official-amd-r9-290x-290-owners-club/41580_20#post_24854355
I've only tested in Time Spy with the Fury X so far as I just got it into my loop yesterday evening, but it seems like I got a decent sample from XFX once again with it clocking at 1160/560 for Time Spy:
The Fury X picked up 8.7% in GS with the overclock of 10.5% on the core and 12% on the HBM.
When compared to one of my best runs with the 480: http://www.3dmark.com/compare/spy/484106/spy/463734#
This is roughly where I envisioned the Fury X to be in relation to my 290 with my gaming clocks of 1200/1500...about a 25% improvement. Will have to do some more testing, but I think this is a good indication of the overall performance boost I will experience. If anything it will probably exceed that value as I game at 5760*1080 and Fiji gets better with higher resolution.