Richland & Kabini rumours

[Nemesis 1]

http://techreport.com/news/24186/new-details-early-benchmark-data-revealed-for-15w-kabini-apu

Exactly Thats why I haven't said one word about 7watt IVB . To me its a none event . But one hell of an accomplishment by intel . The real story the one being ignored because you can't do what you just did is haswell 2 core SoC /soix @7 watts . This 7 watt ivb isn't scarring anybody . Haswell most definit;ly is . or we wouldn't be talking about a short runn product intend to be a stop gap.Same as Hondo . IVB at 7 watts scares no one. hence the reason its talked about rather than haswell. From the link it didn't help your cause add in the 4 watts and its still under 15 watts not that it matters

 

[inf64]

BTW I've just ran that Musemage benchmark on my Radeon HD6870 and got 12300pts score . These ULV products from intel and amd still look like 486 level compared to desktop stuff(and my gfx card is getting old too).

 

[Homeles]

BTW I've just ran that Musemage benchmark on my Radeon HD6870 and got 12300pts score . These ULV products from intel and amd still look like 486 level compared to desktop stuff(and my gfx card is getting old too).

Without knowing the details of how the program assigns scores, it's a bit silly to draw such conclusions.

 

[piesquared]

BTW I've just ran that Musemage benchmark on my Radeon HD6870 and got 12300pts score . These ULV products from intel and amd still look like 486 level compared to desktop stuff(and my gfx card is getting old too).

Yeah, discrete is always going to be higher performing than integrated. I'm looking forward to hearing more about the rumors of 'special sauce' between Kabini and discrete GCN. Like AMD said at their press conference, GCN/GCN +n is key to their strategy going forward.

 

[inf64]

Without knowing the details of how the program assigns scores, it's a bit silly to draw such conclusions.

Single threaded workload (if you meant x86 cores). GPU is fully loaded during the run though. Also, changing CPU frequency brings very small changes to the end score(15% CPU clock increase= 1.5-2% end score increase in my case). So the workload is GPU bound.

 

[Abwx]

One third of the frequency and half the core count and you ll end
6x slower , there s no miracle..

 

[Homeles]

Looks like my point was completely missed. Oh well.

 

[inf64]

Scores vs *cores, got it now . If there is some special formula (a la 3dmark CPU/GPU) then I have not heard about it. Sorry for misunderstanding.
I've found 4600M numbers in Musemage here. 4600M scores 4600pts.

 

[The Alias]

http://www.engadget.com/2013/01/10/amd-interview/
hints to amd being the cpu provider for consoles ? and more specs on temash

 

[piesquared]

Yeah that was a great interview and a few good pieces of information. He makes a lot of good points. I'm wondering though, is this Jolly Green fella typical of tech media???

 

[MightyMalus]

YOU don't know this at all.

Anandtech said it was the top end mobile/notebook haswell part. There were Intel reps in his video talking about it too.

So, we do "know" that GT3 is not all that great. Heck, did you watch the Temash tablet running the game also? Also, it was a demo...unlike the Temash clip where someone was actually playing!

 

[piesquared]

So Richland is integrating eyeSight technology for AMD Gesture Control. #Awesomesauce .

By running the video processing algorithms on the upcoming APUs, eyeSight was able to reach remarkably low CPU consumption compared to running on traditional X86 CPUs. In some cases, the processing time has even been reduced by a factor of 20x- allowing eyeSight to introduce additional code and significantly increase recognition accuracy.

http://www.businesswire.com/news/ho...ing-AMD-APUs-Feature-Built-in-Gesture-Control

 

[Nemesis 1]

So Richland is integrating eyeSight technology for AMD Gesture Control. #Awesomesauce .



http://www.businesswire.com/news/ho...ing-AMD-APUs-Feature-Built-in-Gesture-Control

Well they pretty much had to or be left behind

 

[MightyMalus]

He only has something to say when he is not being quoted.
*ignores*

 

[piesquared]

Well they pretty much had to or be left behind

I wasn't alluding to the fact that they have gesture recognition, that much was known. It's more interesting the massive benefit eyeSight gained from using the new video processing algorithms.

 

[Nemesis 1]

Anandtech said it was the top end mobile/notebook haswell part. There were Intel reps in his video talking about it too.

So, we do "know" that GT3 is not all that great. Heck, did you watch the Temash tablet running the game also? Also, it was a demo...unlike the Temash clip where someone was actually playing!

I clearly stated that I doubted it was SOC haswell. I have seen Soc haswell demoed with GT3 graphics it run fine at 8 watts , I at no time have used GT3e in any of my post. I clearly stated it doesn't matter what it is . Lets see amd below 10 watts running . Intels looks good , Seems to me the ultras are going to have alot of nice choices.

I suppose now if a demo uses the option chip GT3e somehow its cheating , SO we know GT3 isn't that great . Looked dang good to me but intel cheated the used GT3e. If it came on desktop I would get the GT3e I don't have the luxury as of yet,

 

[Nemesis 1]

I forgot to add. The ultras will have a choice of GT3 or GT3e. Why on earth do you think getting to 7 watts is such a big deal to intel . So they could go in tablets , Thats just a bonus.

 

[MightyMalus]

I clearly stated that I doubted it was SOC haswell. I have seen Soc haswell demoed with GT3 graphics it run fine at 8 watts , I at no time have used GT3e in any of my post. I clearly stated it doesn't matter what it is . Lets see amd below 10 watts running . Intels looks good , Seems to me the ultras are going to have alot of nice choices. I suppose now if a demo uses the option chip GT3e somehow its cheating , SO we know GT3 isn't that great . Looked dang good to me but intel cheated the used GT3e. If it came on desktop I would get the GT3e I don't have the luxury as of yet,

There is no GT3 on the desktop chips. And AMD is running below 10 watts...
Temash, hello? GT3 looked fine to me, but not impressive.

 

[Nemesis 1]

Yeah, discrete is always going to be higher performing than integrated. I'm looking forward to hearing more about the rumors of 'special sauce' between Kabini and discrete GCN. Like AMD said at their press conference, GCN/GCN +n is key to their strategy going forward.

This is one I been waiting on also . If AMD sends intel products to test their gpus after they get everthing on the same page I will scream

 

[Nemesis 1]

There is no GT3 on the desktop chips. And AMD is running below 10 watts...
Temash, hello? GT3 looked fine to me, but not impressive.

I didn't look in the box would amd lie . yes they would . I confess i did miss the temash demo . Got a link

 

[Nemesis 1]

There is no GT3 on the desktop chips. And AMD is running below 10 watts...
Temash, hello? GT3 looked fine to me, but not impressive.

I know that or I think I do . Referring to Gt3 on desktop . Intel hasn't said one word about the higher TDC desktop part no one is talking about cause its only been hinted.

It was running along side a NV 650m it looked dang good

 

[mikk]

Thanks for the link. Very interesting detail about the PCH TDP there. So it's 17+4.1W total power for intel IB ULV parts.

Incorrect. ULV chipset has a 3W TDP.

So, we do "know" that GT3 is not all that great.

How can you say that without having fps numbers? With that logic we could say GT650M is not that great, and this chip is 2x the performance of A10-4600M. Richland with just +20% can't reach that performance level.

 

[Nemesis 1]

There is no GT3 on the desktop chips. And AMD is running below 10 watts...
Temash, hello? GT3 looked fine to me, but not impressive.

Ya I just watched the demo again . It does look OK . This tablet space is interesting . ultra space looks locked up with haswell and its options. . Intel never seizes to amaze they found away to make money on there IGPU. Good for you intel , Edram isn't free , and intels charging for it , AMD sleeping again. I how ever seen no proof temash was running at 10 watts . I guess we wait for battery test . Arm chip gave it all away when Anand found out what the wattage spiked to . Let me tell ya . IT was ugly

 

[Nemesis 1]

Incorrect. ULV chipset has a 3W TDP.



How can you say that without having fps numbers? With that logic we could say GT650M is not that great, and this chip is 2x the performance of A10-4600M. Richland with just +20% can't reach that performance level.

Don't debate those 2 or you be on same list with me . The top one never gets info correct until after the fact . The other seems to be a newbe cut him slack

 

[Phynaz]

Kabini benchmarks from Techreport
http://techreport.com/news/24186/new-details-early-benchmark-data-revealed-for-15w-kabini-apu