I have spent the last month benchmarking a selection of AMD and NVIDIA cards with the latest drivers on one of our new 6700k test beds. We are using the AMD Crimson Edition Display Driver, Version 16.8.1 and Nvidia ForceWare 368.81 driver.
We list each resolution test for every game on its own page – meaning if you are just interested in 1440p resolutions for instance, you can skip the other resolutions without effort. If you want to read the whole review and find all the page changes annoying – click on our menu system top right of these pages, and head to ’32. view all pages’.
We are aware the RX 470 is not designed for Ultra HD 4k gaming, but we include the results out of the sake of being thorough. There is also the possibility that if AMD do release an 8GB version of the RX 470, we will want to test it against the 4GB version at higher resolutions. Even if it seems somewhat pointless in the grand scheme of things.
We are using a custom Titan Bayonet system supplied by Overclockers UK as the basis of our test system today. Read more on this system over HERE.
Case: Phanteks Enthoo Evolv ATX Mid Tower
Processor: Intel 6700K @ 4.4ghz
Memory: Corsair Vengeance LPX 16GB (2x8GB) @ 3000mhz
Motherboard: ASUS Z170-E DDR4 ATX Motherboard
Power Supply: Super Flower Leadex 850W Gold Certified
Software: Microsoft Windows 10 64 Bit
SSD: Samsung 250GB 850 EVO
HDD: Seagate 1TB 7,200 rpm 64MB Cache.
If you want to purchase this system yourself head to THIS page on OCUK.
Graphics cards:
Sapphire RX 470 Nitro + OC 4GB – (1,260 mhz core / 1750 mhz (7 Gbps effective) memory)
Comparison Cards on test:
Asus RX 470 Strix Gaming OC Aura RGB – (1,270 mhz core / 1650 mhz (6.6Gbps effective) memory)
Asus RX 480 Strix Gaming OC AURA RGB – (1,330mhz core / 2,000 (8Gbps effective) memory)
Sapphire RX 480 Nitro + OC 8GB – (1,342mhz core / 2,000 mhz (8Gbps effective) memory)
Sapphire RX 480 Nitro + OC 4GB – (1,306mhz core / 1,750 mhz (8Gbps effective) memory)
MSI GTX 1080 Gaming X 8G RGB SLI– (1,683mhz core / 1,823 mhz boost / 5000 mhz memory)
Palit GTX 1080 GameRock Premium Edition – (1,747mhz core / 1,886mhz boost / 5256 mhz memory)
MSI GTX 1080 Gaming X 8G RGB – OC Mode (1,709mhz core / 1,849mhz boost / 5,056 mhz memory)
Gigabyte GTX 1080 G1 Gaming RGB (1721 mhz core / 1860 mhz boost / 5,005 mhz memory)
Nvidia GTX 1080 Founders Edition (1607mhz core/ 1733mhz boost / 5,005 mhz memory)
Nvidia GTX 1070 Founders Edition (1506mhz core/ 1683mhz boost / 4,006 mhz memory)
Sapphire R9 390 Nitro 8GB (Rev 2 w/ backplate). (1040mhz core / 1,500 mhz memory)
Sapphire R9 295X2 (1,018 mhz core / 1,250mhz memory)
AMD R9 Fury X (1,050 mhz core / 500 mhz memory)
AMD R9 Nano (1,000mhz core / 500 mhz memory)
Gigabyte GTX980 Ti XTREME Gaming (1,216 mhz core / 1800mhz memory)
Nvidia GTX Titan Z (706 mhz core / 1,753 mhz memory)
Nvidia GTX Titan X (1,000 mhz core / 1,753 mhz memory)
Asus GTX980 Strix (1,178 mhz core / 1,753 mhz memory)
Nvidia GTX980 Ti (1,000 mhz core / 1,753 mhz memory)
Sapphire R9 390X Tri-X 8GB (1,055 mhz core / 1,500 mhz memory)
Sapphire R9 390 Nitro 8GB (1,010 mhz core / 1,500 mhz memory)
Software:
Windows 10 64 bit
Unigine Heaven Benchmark
3DMark 11
3DMark
Fraps Professional
Steam Client
FurMark
Games:
Ashes Of the Singularity
Dirt Rally
Hitman 2016
Middle Earth: Shadow Of Mordor
Rise Of the Tomb Raider
Grand Theft Auto 5
Metro Last Light Redux
Additional equipment:
Leica S-E (006) medium format camera with 70mm F2.5 and 100mm F2 Leica lens.
We perform under real world conditions, meaning KitGuru tests games across five closely matched runs and then average out the results to get an accurate median figure. If we use scripted benchmarks, they are mentioned on the relevant page.
Game descriptions edited with courtesy from Wikipedia.
<<rw. ★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★✫★★::::::!ir273m:….,….
..
A test on [email protected] doesn’t tell me anything about the card. I don’t know how it acts with i5 2400 or any similar cpu, I don’t know if with this cpu it is much faster than GTX960 and I don’t know how much faster would GTX1060 be this case.
*spoiler* Your fps will be lower, depending on the game, with a weaker cpu. This test is designed to show you the full power of the card because that’s how card comparisons are made, not “But this card does better with a weaker cpu” though there are tests for that */end spoiler*
Next you’re going to ask them to test at 1680×1050?
A simple search led me to a video showing your CPU overclocked to 3.6GHz with a gtx 970 running rise of the tomb raider not bottlenecking it, Witcher 3 bottlenecking the 970 nicely, Crysis 3 shows your 2400 crying silicon tears of pain while the 970 sips electrical tea, and GTA V showing off the godlike optimization of the game, but your CPU running at higher usage than the 970 on average. Now add the performance deltas between the cards in question and the 970 and that’s how much more you’re getting/not getting because your CPU is borderline at 1080p
Wow, 1400Mhz OC…that’s impressive
The pricing is only confusing in the UK, in America it’s not confusing at all….and if it was, that still wouldn’t be as big of a point of contention as the author is making it seem. I feel as though we’ve encountered yet another example of reviewers constantly holding AMD to a higher standard than pretty much every other company. The card is excellent, Sapphire has always made excellent cards, and I feel as though the author is complaining either for the sake of it, or because reviewers just cannot let AMD escape untouched. Seriously, they’re never as hard on Nvidia (or any other company) as reviewers are on AMD…and I’m not even an AMD fan and it’s pretty evident.
Perhaps a word of warning to some potential buyers.
I purchased two Sapphire RX 470 OC+ cards two days ago. My intention was to run them as OpenCL compute devices in a system with other GPUs. The host computer, a fairly new latest-gen APU on a Gigabyte board [88 chipset] was perfectly stable in other configurations {GTX1070-GTX970, GTX970-GTX1070-FX470[XFX card], GTX970-GTX1070-XFX RX470-Integrated Graphics} but It failed to post as soon as I installed the first Sapphire FX470.
Over the course of a couple hours, I tried various configurations, frequently being forced to reset my BIOs to Post. My general conclusion at this point is that the firmware on this card is buggy. It’s fairly new so there are no updates posted.
The card has so far proved to be very unstable in configurations when any other GPU is present, including active integrated graphics. This included my attempts to reliably use it with only one Sapphire RX470 in my primary PCIe v3 x16 slot and a second Sapphire FX470 in my second PCIe v2 x16 slot with APU graphics disabled.
I’m going to give them one more shot with a clean install of Windows but I’m not holding out much hope.
This might not be an issue for you depending on your config but if the cards do, in fact, have buggy firmware that affects their stability in one way, I would be concerned that other issues would also eventually surface.
Otherwise, I can say that from a compute standpoint, the other OC’ed XFX RX470 card I have performs on par with my reference design RX480 card but it costs less and they’re far more available these days. On the down side, the OC versions seem to have their voltage tweaked so their power consumption is actually a bit higher than my stock RX480. The Sapphire 470OC+ has a 8-pin PCIe power header while the standard 480 has a 6pin.
That Wattman Screenshot is actually really useless unless you upload it in a resolution in which we actually can read the numbers 😀
Hi, I understand that you have both the Sapphire Nitro+ RX470 and the XFX RX470…
Which, according to you, is the quieter?
And which is the cooler?
so, would you mind uploading the Wattman settings in higher resolution?
that also means you’re an idiot
Go for Sapphire Nitro+ RX470. The build quality is very superb. I’ve the 8GB version of Sapphire Nitro+ RX470 and it look very nice.
anyone can tell me what a chipset on rx 470 4gb?
hi nvidiot!
The RX470, RX480 or RX570, RX580 all use AMD Polaris Architecture. Not sure if this will help.