Quick News

AMD 300 series launch! Or whatever they will call it!

Not confirmed date, but we already know it will be before end of June(according to Lisa T. Su CEO of AMD) so these rumors appear to be pretty valid.

Previous rumor was launch at Computex, but seems to have changed to E3. E3 is June 16-18, but the reveal is rumored to be June 18. June 18th is the first ever PC Gaming Event at E3, which AMD is a sponser of. So it's fitting to reveal their newest cards on such a grand day for the us PC gamers! June 18th is rumored to be the reveal/launch of the R9 380, R9 370 and R7 360 graphics cards series. On June 18th, they will be available online for everyone. June 24th however, the flagship products – the AMD Radeon R9 390-series will be available online along with a reveal. The Fiji, or R9 390 series, is rumored for GCN 1.3, 4096 cores, 4/8GB of HBM memory(4 for 390, 8 for 390x) and a 1024bit I/O data bus and a ridiculous 640GB/s bandwidth limit. As for the other cards, rumors are they are rebrands but they have all been tweaked and improved upon. Some rumors suggest a tweak to GCN 1.2/1.3 so they can all support freesync and other technologies AMD currently have.

Sources: Source 1
Source 2


It would be interesting to see what the 380 etc perform like but apparently they are just rebranded cards from the 200 series which were rebrands of the 7000 series.

Quite disappointing that what was the 7000 series cards seem to be getting rebranded each year and released as new cards, although Nvidia do the same but seriously it's getting kind of annoying now.

Source
 
Gigabyte to go green again

Gigabyte announced a new mobo for all the Nvidia's "screaming for green again" boyz.

Based on the 1150 socket and Z97 chipset, the brand new Z97X-Game Plus features a black and green theme, and everything you can expect from such a board, at a price of around 150$.

Source
 
Titan X owners get The Witcher III free

Good news for Titan X owners. No matter which retailer you bought your Titan X from, and even if the card isn't technically included amongst the card that comes with the current game bundle, Nvidia has decided to grant Titan X owners a copy of The Witcher III.

Just redeemed the game my self to confirm and it is working.

If you don't get the reward pop up. Just uninstall Geforce Experience and install it again and it should pop up. Then simply follow the instructions to have the game added to your GOG.com account.
 
Some of the AMD 300 series NDA was lifted today, first to report on some confirmed details from the embargo lifting was TweakTown.. needless to say it's all interesting stuff on HBM memory. 94% less space taken up by HBM than GDDR5 along with more bandwidth and power efficiency? Yes please;)

Only thing that worries me though is the 1Gbps clockspeed.. or 500Mhz. Wonder if that effects anything else even though it's far far faster architecturally? Some games may not like it? We'll see i guess.

Source: http://www.tweaktown.com/articles/7...cards-starting-amds-radeon-r9-390x/index.html
 
Last edited:
Some of the AMD 300 series NDA was lifted today, first to report on some confirmed details from the embargo lifting was TweakTown.. needless to say it's all interesting stuff on HBM memory. 94% less space taken up by HBM than GDDR5 along with more bandwidth and power efficiency? Yes please;)

Only thing that worries me though is the 1Gbps clockspeed.. or 500Mhz. Wonder if that effects anything else even though it's far far faster architecturally? Some games may not like it? We'll see i guess.

Source: http://www.tweaktown.com/articles/7...cards-starting-amds-radeon-r9-390x/index.html

Downside to HBM v1 is its also limited to maximum 4gb memory, but can potentially give you a 4096bit wide IO. 1024 bit per package. Might be a downside but still a huge upside to the technology! Not bad for the first version right?

GDDR5 - Up to 28Gb/s per chip
HBM - >100Gb/s per stack
 
looks like it is limited to 4gb or vram for the time being though

4gb is still plenty for me, however what I don't like the idea of, is based on some diagrams, this stacked memory allows the HBM to be placed much, much closer to the GPU die allowing reduced circuitry and reduced latency etc. When it comes to watercooling blocks, I think things will become more fragile considering the IC stacks will be closer and higher than the GPU die. I see potential broken chips from installing waterblocks for the first time. Maybe its nothing, but we will see when the 390x comes.

I think AMD have hit it big time here with the possibilities they provide.

edit*

actually ignore all this. Apparently AMD confirmed the stack will not be higher than the GPU. They have just blown things out of proportion to explain the technology in detail. In practise the stack heights will not be anything to worry about
 
Last edited:
4gb is still plenty for me, however what I don't like the idea of, is based on some diagrams, this stacked memory allows the HBM to be placed much, much closer to the GPU die allowing reduced circuitry and reduced latency etc. When it comes to watercooling blocks, I think things will become more fragile considering the IC stacks will be closer and higher than the GPU die. I see potential broken chips from installing waterblocks for the first time. Maybe its nothing, but we will see when the 390x comes.

I think AMD have hit it big time here with the possibilities they provide.

edit*

actually ignore all this. Apparently AMD confirmed the stack will not be higher than the GPU. They have just blown things out of proportion to explain the technology in detail. In practise the stack heights will not be anything to worry about

aye 4gb is still plenty for most at the moment, i wonder how these chips an memory are going to perform under water. well we wont have to wait too long now to find out i guess
 
Where are you getting the 4GB limit from??? As it is different from what I have been reading for months.
 
Where are you getting the 4GB limit from??? As it is different from what I have been reading for months.

Taken from Guru3D who did an article on HBM

1st gen HBM is limited to 4GB Graphics memory

These stacked memory packages have limitations, in the first generation you are looking at four stacks per package with two 128MB chunks in each layer, so that is 256MB per layer. Four times 256MB = 1024MB (or 1 Gigabyte) per accumulated DRAM stack/package. Currently chip designs allow for four stacks per IC (GPU). So that is 4 packages x (4x256MB) = 4096 MB. Hence the one limitation (if you can call it that) is HBA memory being limit at a maximum 4GB of graphics memory for the graphics card.

But is it 1024-bit or 4096-Bit? Here we'll need to 'zoom in' towards one DRAM layer first. Each layer is has two memory parts each at 128-bit meaning HBM is using 128-bit wide channels so eight eight of allow for a full 1024-bit interface. Total bandwidth is in the 128GB/s range with die stacks of four DRAM dies. Important to know is that each memory controller is independently timed and controlled. So 256-bit x 4 = 1024-bit per package. If an SoC/Processor/GPU/APU is fitted with four stacked packages then that would boil down to 4096-bit (wide IO). HBA is in this sense completely different from GDDR5

Source
 
There are ways around the 4GB limitation however. But we can't get off topic.. this is a quick news thread after all, If you want to discuss it more in detail than start a thread:p
 
Back
Top