AMD: Speculation, Rumors, and Discussion (Archive)

Status
Not open for further replies.
Depends on the competitive landscape, doesn't it? HD 4770 as a pipe cleaner for 40 nm was a smart move in my books.
 
Wait, clamshell mode generally consumes less power? Or are you saying it's less than twice the power from the chips? Very neat detail!
As others have said already, a single chip consumes less power when in clamshell (x16) mode compared to the "normal" (x32) mode. Just checked the maximum specified power consumption of a single H5GQ1H24AFR-R0C GDDR5 memory device from SKHynix (6Gbps). It's 3.02W in x32 mode (achieved during read bursts), but 1.73W during reads in x16 mode. Somewhat peculiar, the write burst consumption doesn't drop as much (from 2.91W to 1.86W, which marks the maximum in x16 mode). And while the absolute numbers for the power consumption changes with speed grade, process, supply voltage, whatever, I would guess the ratio between x32 and x16 mode not to be influenced too much by it.

In any case, doubling memory capacity by using the clamshell mode and doubling the number of chips, increases the power consumption of the memory system not by a huge amount. And it's conceivably even less of an increase when doubling the capacity of the chips and not putting twice the amount of memory devices on the PCB of the card.
 
Chips that good undermine the theory of pipe cleaners.
Indeed. I had it and the 40nm HD4770 performed admirably compared to the 55nm HD4870 it replaced at far lower power draw and noise.
Polaris 10 would seem to have a similar relationship to Hawaii, but even more pronounced. Small wonder though, GPUs on the 28nm node has been going for almost 4.5 years now.
 
As others have said already, a single chip consumes less power when in clamshell (x16) mode compared to the "normal" (x32) mode. Just checked the maximum specified power consumption of a single H5GQ1H24AFR-R0C GDDR5 memory device from SKHynix (6Gbps). It's 3.02W in x32 mode (achieved during read bursts), but 1.73W during reads in x16 mode. Somewhat peculiar, the write burst consumption doesn't drop as much (from 2.91W to 1.86W, which marks the maximum in x16 mode). And while the absolute numbers for the power consumption changes with speed grade, process, supply voltage, whatever, I would guess the ratio between x32 and x16 mode not to be influenced too much by it.

In any case, doubling memory capacity by using the clamshell mode and doubling the number of chips, increases the power consumption of the memory system not by a huge amount. And it's conceivably even less of an increase when doubling the capacity of the chips and not putting twice the amount of memory devices on the PCB of the card.
Very neat. Where do you find spec sheets for memory chips like that?
 
The hype are in most part made by AMD fans themselves. But i will find disappointing if Rx480 don't manages to match a 390x at 1080p, at least.
 
The hype are in most part made by AMD fans themselves. But i will find disappointing if Rx480 don't manages to match a 390x at 1080p, at least.

390 / 390x performance at around a 100w and $200 bucks is not a bad deal tho. That's the point of the card. The 1070 and 1080 are great cards but very expensive. I can buy one of these Rx480s and have the same exact or better experience on my gfs pc that I have on mine running a 290 non x .

Heck if this hits 390x performance I will prob sell or give away my card and get this since it will use less power , create less heat and if someone makes a bracket to hold my aio water cooler , it will be whisper silent
 
Haha, I gotcha. You recited the exact model number of the memory chip, so I assumed a deeper understanding.
As said, I remembered seeing a datasheet a while back and the relative power consumption of the x32 and x16 modes in there (I didn't remember the exact numbers after a two or three years of course) and was confident enough to post this. After you expressed some interest I searched for some datasheet and posted the specific numbers for a specific memory chip (which is the correct way to do as they will vary a bit from model to model). And what "deeper understanding" did you expect? The reason for this was given already (simply less bits transferred at high speeds per chip).
To make it short: I don't get your "Haha, I gotcha". You couldn't possibly get me, just some information. ;)

Anyway, I guess it got too OT.

PS: "Get" has far to many meanings in English.
 
Last edited:
390 / 390x performance at around a 100w and $200 bucks is not a bad deal tho. That's the point of the card. The 1070 and 1080 are great cards but very expensive. I can buy one of these Rx480s and have the same exact or better experience on my gfs pc that I have on mine running a 290 non x .

Heck if this hits 390x performance I will prob sell or give away my card and get this since it will use less power , create less heat and if someone makes a bracket to hold my aio water cooler , it will be whisper silent
Where did this 100 watt figure come from in the first place? It's quite a bit below the 150 watts that was given earlier, isn't it?
 
the 150 came from an AMD chart that said "power" only, not tdp. But when they said 150w, they meant it in the context that it would only need one 6 pin header (pci-e 75w + 6 pin 75w = 150). They also wouldn't release a card that at stock would consume 100% of the power available to them (150w), that would make the card out of spec (and needing an 8 ping instead of 6 to be compliant)
 
Status
Not open for further replies.
Back
Top