Corsair QL120 and QL140 RGB fan review
Promo: Windows 10 Pro for $13 With Office 2016 For $33
Corsair Void RGB Elite Wireless Headset review
Team Group PD400 Portable SSD review
AMD Athlon 3000G review
Team Group T-Force Delta Max 1 TB SSD review
Guru3D Rig of the Month - November 2019
ASUS ROG Rampage VI Extreme Encore review
Toshiba RC500 500GB NVMe M.2 SSD review
Promo: URcdkey Black Friday Windows 10 Pro for $12
AMD Releases Radeon R9 Fury X details + exclusive photos
Okay, I have the two thumbs up to share the official data with you guys. AMD will release its high-end card based on FIJI GPUs very soon, expect reviews next week as well as availability. In this news-item I am going to share the official details and specifications of the Fury X alongside some photos.
Fiji much like Tonga makes use of 3rd-gen GCN 1.2 architecture, same architecture but obviously scaled up towards a huge 4096 shader processors. The GPU core will be clocked at 1050 MHz and is tied to 4 GB High Bandwidth Memory. This memory runs at a nice 512 GB/s thanks to the extremely wide 4096-bit memory bus. based on 64 compute units thats 64 shader units per CU cluster. It has 64 ROP units and a whopping 256 texture memory units. The product will deliver performance of 8.6 TFLOPS. The HBM memory is indeed 4GB in total, four stacks of 1GB being clocked at 500 MHz, combined they will deliver 512 GB/s of bandwidth.
The power usage will sit at roughly 275 Watt. The product as you guys know is liquid-cooled. Fury X will get three DP 1.2 ports and one HDMI 1.4 connector (there is no HDMI 2.0 support). The Fure X will be available starting June 24th for a price of 649 USD, which is 700 to 750 EURO based on your country and VAT. I attached some screen-shots that display all relevant information, have a peek please.
There she is, the LEDs are power load indicators, at this fully lit stage the GPU is 100% utilized.
Everything about this card screams 4K. Below I have photos of the die and product I took yesterday at an AMD event. Again we have a unit in-house and you can expect the full review next week.
AMD releases list of compatible DirectX 12 graphics cards & APUs - 05/14/2015 01:02 PM
AMD posted a thing or two on its blog about the DX12 API. The blog entry also lists a handy overview of what cards are actually DX12 compatible....
AMD releases The Fixer 3 video - 01/30/2015 04:46 PM
AMD just submitted the new Fixer Video to us to share with you guys. When the enemy is at the gates, there’s only one person to call. But you don’t have to call him, he shows up, i...
AMD releases new 12- and 16-Core Opteron 6300 Series Processors - 01/22/2014 09:16 AM
AMD today announced the immediate availability of its new 12- and 16-core AMD Opteron 6300 Series server processors, code named "Warsaw." Designed for enterprise workloads, the new AMD Opt...
AMD Releases Catalyst 12.1 WHQL Driver Download - 01/26/2012 12:29 PM
AMD has made available the first WHQL-certified graphics drivers of 2012, the Catalyst 12.1. This release includes support for Radeon HD 2000, HD 3000, HD 4000, HD 5000 and HD 6000 series desktops (an...
AMD releases Catalyst 11.11c performance driver - 12/02/2011 12:47 PM
AMD Catalyst 11.11c performance driver features Elder Scrolls Skyrim New in Catalyst 11.11c: Delivers AMD CrossfireX performance scaling for the AMD Radeon HD 5000 Series Delivers AMD CrossfireX pe...
Hirantha
Senior Member
Posts: 254
Joined: 2012-06-26
Senior Member
Posts: 254
Joined: 2012-06-26
#5099737 Posted on: 06/18/2015 02:48 PM
why so many different names? soooo confusing.
why so many different names? soooo confusing.
snip3r_3
Senior Member
Posts: 2985
Joined: 2004-12-17
Senior Member
Posts: 2985
Joined: 2004-12-17
#5099749 Posted on: 06/18/2015 02:53 PM
Fury series is the brand new "Fiji" GPU. The 300 series are simply all rebrands.
why so many different names? soooo confusing.
Fury series is the brand new "Fiji" GPU. The 300 series are simply all rebrands.
ruiner13
Senior Member
Posts: 145
Joined: 2013-11-22
Senior Member
Posts: 145
Joined: 2013-11-22
#5099753 Posted on: 06/18/2015 02:56 PM
I like the design, but really no HDMI 2.0? and 275W at stock clocks is... well... meh. This reeks of being rushed to market. They could have waited for a die shrink and HBM2 for 4+ GB, but it seems the 980Ti really forced their hand, if you ask me.
I like the design, but really no HDMI 2.0? and 275W at stock clocks is... well... meh. This reeks of being rushed to market. They could have waited for a die shrink and HBM2 for 4+ GB, but it seems the 980Ti really forced their hand, if you ask me.
vbetts
Moderator
Posts: 14715
Joined: 2006-07-04
Moderator
Posts: 14715
Joined: 2006-07-04
#5099765 Posted on: 06/18/2015 03:01 PM
Even if the performance is not top of the charts, it more or less is a proof of concept to show that it does have its advantages.
This product now is out there just to show that it's possible to have this tech in a consumer setting. It's AMD saying "Hey, it works and we will improve the next revision."
I like the design, but really no HDMI 2.0? and 275W at stock clocks is... well... meh. This reeks of being rushed to market. They could have waited for a die shrink and HBM2 for 4+ GB, but it seems the 980Ti really forced their hand, if you ask me.
Even if the performance is not top of the charts, it more or less is a proof of concept to show that it does have its advantages.
This product now is out there just to show that it's possible to have this tech in a consumer setting. It's AMD saying "Hey, it works and we will improve the next revision."
PhazeDelta1
Moderator
Posts: 15616
Joined: 2010-09-12
Moderator
Posts: 15616
Joined: 2010-09-12
#5099771 Posted on: 06/18/2015 03:04 PM
Well at least AMD did one thing right with this card. They put a Nidec Servo fan on the raidiator.
Well at least AMD did one thing right with this card. They put a Nidec Servo fan on the raidiator.
Kyrat
Member
Posts: 74
Joined: 2014-11-07
Member
Posts: 74
Joined: 2014-11-07
#5099784 Posted on: 06/18/2015 03:22 PM
What is this torz thing?
What is this torz thing?
Lowki
Senior Member
Posts: 622
Joined: 2011-11-26
Senior Member
Posts: 622
Joined: 2011-11-26
#5099786 Posted on: 06/18/2015 03:24 PM
Cant wait for the review.
Cant wait for the review.
TheSeekingOne
Junior Member
Posts: 10
Joined: 2015-05-07
Junior Member
Posts: 10
Joined: 2015-05-07
#5099787 Posted on: 06/18/2015 03:25 PM
Top tear performance, H.265 video decode accelerator, very decent FP64 math support, decent DSP for audio processing, half a terabyte of low latency memory bandwidth and all that in a 275W package.
This is a high-end very-well-engineered product.
Well at least AMD did one thing right with this card. They put a Nidec Servo fan on the raidiator.
Top tear performance, H.265 video decode accelerator, very decent FP64 math support, decent DSP for audio processing, half a terabyte of low latency memory bandwidth and all that in a 275W package.
This is a high-end very-well-engineered product.
Sukovsky
Senior Member
Posts: 915
Joined: 2005-08-22
Senior Member
Posts: 915
Joined: 2005-08-22
#5099796 Posted on: 06/18/2015 03:33 PM
Getting a Fury X. Sold my 970 for the price I bought it for, lol. Can't wait, my Intel HD4000 isn't exactly a party.
Getting a Fury X. Sold my 970 for the price I bought it for, lol. Can't wait, my Intel HD4000 isn't exactly a party.
StarvinMarvinDK
Senior Member
Posts: 1338
Joined: 2004-12-19
Senior Member
Posts: 1338
Joined: 2004-12-19
#5099811 Posted on: 06/18/2015 03:47 PM
Not significantly better for new tech...
Not significantly better for new tech...
cyclone3d
Senior Member
Posts: 419
Joined: 2006-11-10
Senior Member
Posts: 419
Joined: 2006-11-10
#5099828 Posted on: 06/18/2015 04:06 PM
Although I am not planning on upgrading my video card setup any time soon, I can't wait to see what this card can do.
The latency on the new HBM RAM setup has to be crazy low.
The throughput of the RAM is higher than the L2 cache on an Intel 4960x running at stock speeds.
You have a tiny board compared to older high end cards.
Although I am not planning on upgrading my video card setup any time soon, I can't wait to see what this card can do.
The latency on the new HBM RAM setup has to be crazy low.
The throughput of the RAM is higher than the L2 cache on an Intel 4960x running at stock speeds.
You have a tiny board compared to older high end cards.
Picolete
Senior Member
Posts: 277
Joined: 2014-12-09
Senior Member
Posts: 277
Joined: 2014-12-09
#5099831 Posted on: 06/18/2015 04:10 PM
Not significantly better for new tech...
]
Why would anyone use AA in 4k?
Not significantly better for new tech...
]
Why would anyone use AA in 4k?
weasel
Senior Member
Posts: 676
Joined: 2013-04-17
Senior Member
Posts: 676
Joined: 2013-04-17
#5099849 Posted on: 06/18/2015 04:21 PM
Looks nice like the led power indication
Looks nice like the led power indication
schmidtbag
Senior Member
Posts: 4568
Joined: 2012-11-10
Senior Member
Posts: 4568
Joined: 2012-11-10
#5099857 Posted on: 06/18/2015 04:26 PM
The way the pipes are routed (from the "right" side) might be a small issue for people with larger tower coolers in a smaller case. Then again, there doesn't seem to be a really ideal way to place it without hitting certain limitations in certain form factors/cases.
GCN 1.2 though, meaning it was indeed a "brute force" design. Can't wait to see the numbers on a first gen HBM implementation though. Being GCN 1.2 it also should allow for some rough extrapolation of just how much GCN (at least 1.2 with some compression technology) was bandwidth constrained in the past vs 285/380's GDDR5.
GCN 1.2 was never really used to it's maximum potential. The R9 285 should have had at least 3GB and the new 380 isn't really any different.
I personally think AMD should have gone this route:
* The Fury GPUs should have been the 390 and 390X
* The updated Hawaii GPUs should have been the 380 and 380X
* The updated GDDR5 Tonga GPU should have been the 370
* An HBM Tonga GPU which would be the 370X
I think if they did things like this, people would be a little more forgiving of all the rebrands.
The way the pipes are routed (from the "right" side) might be a small issue for people with larger tower coolers in a smaller case. Then again, there doesn't seem to be a really ideal way to place it without hitting certain limitations in certain form factors/cases.
GCN 1.2 though, meaning it was indeed a "brute force" design. Can't wait to see the numbers on a first gen HBM implementation though. Being GCN 1.2 it also should allow for some rough extrapolation of just how much GCN (at least 1.2 with some compression technology) was bandwidth constrained in the past vs 285/380's GDDR5.
GCN 1.2 was never really used to it's maximum potential. The R9 285 should have had at least 3GB and the new 380 isn't really any different.
I personally think AMD should have gone this route:
* The Fury GPUs should have been the 390 and 390X
* The updated Hawaii GPUs should have been the 380 and 380X
* The updated GDDR5 Tonga GPU should have been the 370
* An HBM Tonga GPU which would be the 370X
I think if they did things like this, people would be a little more forgiving of all the rebrands.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 2985
Joined: 2004-12-17
The way the pipes are routed (from the "right" side) might be a small issue for people with larger tower coolers in a smaller case. Then again, there doesn't seem to be a really ideal way to place it without hitting certain limitations in certain form factors/cases.
GCN 1.2 though, meaning it was indeed a "brute force" design. Can't wait to see the numbers on a first gen HBM implementation though. Being GCN 1.2 it also should allow for some rough extrapolation of just how much GCN (at least 1.2 with some compression technology) was bandwidth constrained in the past vs 285/380's GDDR5.