[WCCF] AMD Radeon R9 390X Pictured

Page 81 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

cbrunny

Diamond Member
Oct 12, 2007
6,791
406
126
So did they not actually do anything at Computex then? Geeze. Waited too long. I bought speakers instead. Will keep on with my CF 7970s until the cashflow is full again.
 
Feb 19, 2009
10,457
10
76
Difference in games seems unlikely, since two sites using the same game (metro last light) sees 42 and 102 Watts respectively. Also all of the cards should be reference (as far as I can tell), so there shouldn't be that big of a difference in bios settings.

Any way I'm not saying there's anything of note here, I just stumbled over it and found it mildly interesting.

You should check if the card is reference. The reference card does use more power than custom models.

8GB model:
power-consumption4.png


4GB model:
GPU-power-consumption.png


power_average.gif


power_average.gif


They also peak spike less (less leakage running cool than reference at 94C) so if some sites measure peak system load, the difference can be huge.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,476
136
As you well know, AMD has always competed with a smaller die and they come close to matching NV's top GPU for many generations now. The last time they moved first with new vram tech, they stomped on NV hard with a very small GPU that was half the size of NV's.

The performance gap was closed with Hawaii. It shows AMD is able to match NV with a smaller GPU. The next step that is logical, would mean their huge GPU + new vram tech should stomp on NV's huge GPU.

If it doesn't, it's a flop and going backwards in progress.

I mentioned this awhile ago and @RS and some others seem to think its OK for AMD to be slower, as long as they priced it cheaper. NO. Lisa is different, she's fully aware that approach is what has led them to this situation, losing the performance crown and forced to go cheap is NOT a good business strategy. Not for the CPUs and not for GPUs.

It's not OK to go balls to the walls with a massive GPU and new vram tech and be slower
.

I totally agree with you. The objective for going for a monster GPU die (>= 600 sq mm) must be to take the GPU crown unequivocally. Thats the right way. AMD needs to show they can lead the GPU market with the fastest GPU in the world. Anyway I have put down my speculations based on leaks and my guess is perf/sp has gone up significantly with R9 3xx due to further architectural improvements over those seen in R9 285. Specifically I expect AMD to have addressed perf/sp improvements as thats just the logical extension of Tonga improvements.

http://www.overclock.net/t/1558034/r9-r7-3xx-series-architectural-and-performance-speculation

My thoughts are AMD will take the GPU crown with a clean kill (>15%) :D Anyway just 3 weeks before we get the facts on the table. :thumbsup:
 

Azix

Golden Member
Apr 18, 2014
1,438
67
91
Hybrid 390x why ? They must have done something or it's raining water coolers in the graphics industry. Just overclocking it to death isn't great.
 

Glo.

Diamond Member
Apr 25, 2015
5,930
4,991
136
I hope that doesn't mean 3GB of Vram instead of 4?

It can also be a 6 GB card ;).

Wasn't 1792 GCN core chip supposed to have 4 GB of RAM and 256 Bit memory bus? Ultimately named R9 380?
 

LTC8K6

Lifer
Mar 10, 2004
28,520
1,576
126
It can also be a 6 GB card ;).

Wasn't 1792 GCN core chip supposed to have 4 GB of RAM and 256 Bit memory bus? Ultimately named R9 380?

But 6GB would be a waste on a card that can't utilize it.

Not that we haven't seen that before. :)
 

Glo.

Diamond Member
Apr 25, 2015
5,930
4,991
136
So in theory 2048 GCN core and 384 Bit 3/6 GB of VRAM GPU would be... R9 380X.

It makes perfect sence, so far ;).
 

LTC8K6

Lifer
Mar 10, 2004
28,520
1,576
126
So in theory 2048 GCN core and 384 Bit 3/6 GB of VRAM GPU would be... R9 380X.

It makes perfect sence, so far ;).

If 380 has 4gb, then 380x won't have 3...

It will have 6, which it probably can't take advantage of.
 
Feb 19, 2009
10,457
10
76
So in theory 2048 GCN core and 384 Bit 3/6 GB of VRAM GPU would be... R9 380X.

It makes perfect sence, so far ;).

Full Tonga on GloFo may have a good chance competing against 960 due to improved perf/w, but if its still from TSMC, it's not going to change much.
 

antihelten

Golden Member
Feb 2, 2012
1,764
274
126
You should check if the card is reference. The reference card does use more power than custom models.

I know there can be a difference between reference and custom cards, but that's not the issue. The issue is that sites using reference 980s and reference 290Xs, are measuring power deltas between the two cards that variate by as much as 60W.

Interestingly, comparing the 980 to the 980 Ti, the sites seem to be in somewhat better agreement (with 1 or 2 outliers):

50ish Watt group:
Techspot (37W delta - Metro Last Light)
Techreport (40W delta - Crysis 3)
THG (48W delta - says it's a game, but not which one)
TPU (55W - Metro Last Light)
Hardware Canucks (61W delta - Hitman)

computerbase.de (70W delta - Ryse)

100ish Watt group:
Anandtech (87W delta - Crysis 3)

Which would indicate that it is mainly the reference 290X samples review sites received that are variating quite significantly.
 
Feb 19, 2009
10,457
10
76
I know there can be a difference between reference and custom cards, but that's not the issue. The issue is that sites using reference 980s and reference 290Xs, are measuring power deltas between the two cards that variate by as much as 60W.

Interestingly, comparing the 980 to the 980 Ti, the sites seem to be in somewhat better agreement (with 1 or 2 outliers):

50ish Watt group:
Techspot (37W delta - Metro Last Light)
Techreport (40W delta - Crysis 3)
THG (48W delta - says it's a game, but not which one)
TPU (55W - Metro Last Light)
Hardware Canucks (61W delta - Hitman)

computerbase.de (70W delta - Ryse)

100ish Watt group:
Anandtech (87W delta - Crysis 3)

Which would indicate that it is mainly the reference 290X samples review sites received that are variating quite significantly.

You also have to check if they measure peak gaming load or average gaming load. A peak measurement may not reflect accurately, kinda like min/max fps without context. Is it a minor blip or is it constant peaks?

Its best to compare per site since one would expect they repeat the test the same way.
 

Despoiler

Golden Member
Nov 10, 2007
1,968
773
136
So it's 600mm2+ MINUS the memory controller which eats up a alot of die space. So it could have an effective 700-800mm2?

A man can dream.

Also 2GB is more than enough :p

The memory controller is the logic slice that sits under the HBM dram slices. It's no longer part of the GPU die.

HBM_8_TSVs.png
 

antihelten

Golden Member
Feb 2, 2012
1,764
274
126
You also have to check if they measure peak gaming load or average gaming load. A peak measurement may not reflect accurately, kinda like min/max fps without context. Is it a minor blip or is it constant peaks?

Its best to compare per site since one would expect they repeat the test the same way.

Hardware canucks is peak (honestly thought it was average, but on closer inspection this wasn't the case), the rest should be average load (although in some cases it's not entirely clear.

I don't really see what sites repeating their tests has to do with this. it's an issue of inter site variation, not intra site variation (which I would very much hope is no where near 60W).

Anyways this is all getting fairly off topic.
 
Status
Not open for further replies.