the card do 64% of TDPs
The TDP is not getting close to even 100%.
The MSI card I run has a max of 110%.
Interestingly the TDP is only 109 so did not reach the limit.
Is there a base to all thoses % ages.?.
Something like a number in actual watts.?.
Is there a base to all thoses % ages.?.
Something like a number in actual watts.?.
% of the TDP of the 980? Which is 165W.
165W is some cool magic power number
The actual TDP is 180W. Then it's a bit tricky. The power % reported by Afterburner or similar tools is the biggest of those 4 values :
- total 12V power / TDP
- 12Vbus power / 12Vbus power limit (maybe with a bias to take into account 3.3Vbus power)
- 12V1 power / 12V1 power limit
- 12V2 power / 12V2 power limit
With Afterburner you cannot know what the power number means. However GPU-Z always reports the first value while nvidiainspector reports both the first value (called GPU power level) and the biggest of those 4 values (called total power level). Usually with a high load the first value is always the biggest one. At idle or lower GPU load the second value might be the biggest one.
D
Deleted member 2197
Guest
Some tidbits on the 980M and 970M GPU's ...
http://www.guru3d.com/news-story/nvidia-geforce-gtx-980m-gtx-970m.html
A bit of a pleasant surprise to see the mobile parts out so quickly.
I must admit I am a bit surprised at the configuration. 980M is a fairly cut down part with just 12 SMMs, i.e. 75% of the fully enabled GM204. Its clocked fairly high though, almost matching the desktop parts.
This seems to be a departure from the conventional strategy for the mobile sector during the last few years to use fully or almost fully enabled chips at lower clocks. This was touted as the most power efficient solution. Seems to be quite a bit of a U-turn this time around.
I must admit I am a bit surprised at the configuration. 980M is a fairly cut down part with just 12 SMMs, i.e. 75% of the fully enabled GM204. Its clocked fairly high though, almost matching the desktop parts.
This seems to be a departure from the conventional strategy for the mobile sector during the last few years to use fully or almost fully enabled chips at lower clocks. This was touted as the most power efficient solution. Seems to be quite a bit of a U-turn this time around.
Anandtech says 75% but I'm skeptical of that as well. We may see that in some cases, but should be more like 70% in general.
Last edited by a moderator:
70% look effectively spot on due to spec .. ( and this is the number "pre- review today are using for try get an idea of the performance ).
Well the boost clock ist at 1038mhz, this mean it is close to the base clock of the 980 ( official 980 boost clock = 1216mhz ). so nearly 200mhz under. This said with turbo you have a good control by corespeed of the TDP and temps automatically made.. As for disabled chips, this let them a good margin for relase a GTX985M then.
Last edited by a moderator:
Anandtech says base clock is 1038 Mhz, not boost clock. They dont say what the boost clock is. I don't think they would boost as high as the desktop parts though..especially in sustained use cases. Lets wait till a proper review is out before we jump to conclusions though.
Yes..it leaves them room to release a fully enabled or if not, a slightly less cut down GM204 in the future.
The site i have watch was giving it as boost clock ( base is unkown ), who seems logic after all, due to how work the "new" turbo... but you could be right and they was wrong. And like Anand is in general right, well it could be that.
If that 1038Mhz is base clock, 75-80% of (reference) GTX 980 might be doable. The GTX 970 isn't 20% behind GTX 980 after all and would have just about the same base clock (has faster memory than that GTX 980M though). But if the TDP is 120W (and you've really got sufficient cooling, something which isn't always the case in mobile designs) I could see close to 80% of GTX 980 performance as being possible.
I think it's supposed to offer 80% of the GTX 970's performance; which wouldn't be bad at all.
It has only 75% of the CUDA cores and 71% of the memory B/W of GTX 980 to begin with. And base clock is 1038 mhz vs 1126 mhz of the GTX 980. So we're looking at 75% X (1038/1126) = 69% of theoretical GTX980 performance, not considering the fact that it may not boost as much due to TDP and/or cooling constraints. Hence I see ~70% being a realistic number.
Base clock of 970M is 924 mhz so if we extrapolate for GTX980, a 1038 mhz base clock yields 1164 mhz boost clock. Its not too far off the desktop part but it remains to be seen how high the boost clock will go to in sustained workloads in a laptop.
Nope they specifically referenced it to GTX 980.
970M is already using a rather crippled chip so I don't see a further cut down GM204 being used. A hypothetical 10 SMM GM206 seems more likely IMHO.
Well these are meant for gaming laptops so...
Just curious, what games could you run on it which would actually benefit from 3 GB VRAM?
Some tidbits on the 980M and 970M GPU's ...
http://www.guru3d.com/news-story/nvidia-geforce-gtx-980m-gtx-970m.html
Being a half-empty glass kinda guy, this to me says that they didn't push the 980 desktop performance far enough. Given the power consumption and heat dissipation headroom advantage that a desktop GPU has over a mobile GPU, the performance gap should always be big.
