Richland & Kabini rumours

[Eyefinity]

Originally Posted by Evleos

(2) Test conducted in AMD Performance Labs using FutureMark 3DMark Vantage P as a metric for GPU performance and PCMark Vantage as a metric for CPU performance. The AMD Z-60 APU-based system scored 455 and 1914 in 3DMark and PCMark respectively. The AMD A6-1450 APU-based system delivers scores of 981 and 3123 in 3DMark and PCMark respectively. The platforms tested were the AMD Z-60 APU with AMD Radeon(TM) HD 6250 graphics, 2x 2GB DDR3-1066, Microsoft Windows 7 and the "Larne" reference platform with an AMD A6-1450 Quad Core 1.0GHz APU, AMD Radeon(TM) HD 8280 Series graphics, 2GB DDR3-1066 system memory and Microsoft Windows 7. TEM-2

Tablet platforms:

2012: 455 in 3DMark and 1914 in PCMark
2013: 981 in 3DMark and 3123 in PCMark

(3) Test conducted in AMD Performance Labs measuring productivity performance with PCMark Vantage. The "Kabini" A6 APU-based system scored 5271 while the "Brazos" APU-based system scored 2807. "Kabini" PC config is based off the "Larne" reference design with 2013 AMD A6-5200 APU with AMD Radeon HD 8400 graphics, 4G DDR3 1600, and Windows 8 64bit. "Brazos" PC config is based off the "Renmore" reference resign with 2012 AMD E2-1800 APU with AMD Radeon HD 7340 graphics, 4G DDR3 1333 and Windows 7 Ultimate. KBN-3

Netbook/AIO/cheap ultrabook platforms:

2012: 2807 in PCMark Vantage
2013: 5272 in PCMark Vantage

So I guess they're not cherry picking their benchmarks after all, considering how general 3DMark and PCMark are.

Whoah nice.

So this A6-1450 is the top "Temash" quad core, expected TDP ~5W, with double the performance of the Z60 in graphics and some 50%+ in x86.

That means that this ~5W A6 1450 will have performance better than the current 18W E-450. Intel should be quite afraid.

[Evleos]

Originally Posted by Abwx

Thanks for the infos..

That makes CPU/GPU respective improvements for A10 and A8
at 11-21% and 12-40% respectively.

Seems that they focused on portable gears , we ll see if there will
be desktop versions..

Richland = clock bump + larger GPU?

[Nemesis 1]

Originally Posted by Abwx

An official number....

Total CPU and GPU SHE say Cpu or APU she was talking over all . SO that could be 20% +30%= 50% I loved the demos against intels processors .

[Abwx]

Originally Posted by Evleos

Richland = clock bump + larger GPU?

Likely the easiest way to improve the things a little bit ,
still not significantly but a welcomed bump given that it
is made at little if any cost.

Edit : 11-12% frequency bump for the CPU is still
a lot given the thermals of mobile SKUs.
Might well be also an IPC improvement of a 2-5%.

[Abwx]

Originally Posted by Nemesis 1

Total CPU and GPU SHE say Cpu or APU she was talking over all . SO that could be 20% +30%= 50% I loved the demos against intels processors .

20% and 30% would be 25% average improvement , unless
you re using your usual Intel dedicated metric.....

50% average could mean 30% CPU and 70% GPU....

[Olikan]

Originally Posted by Abwx

I think that it s mainly a GPU improvement , but then without a slightly
better CPU it would be diffucult to extract more perfs unless this was only
a matter of bandwith...

Good point, maybe AMD improved it's memory controlers

[Eyefinity]

Who cares about Richland? :p

Temash is faster than Brazos at 1/3rd TDP.

[MightyMalus]

Makes me wish I could use a Temash as my GPU! rofl xD

[antihelten]

Originally Posted by Eyefinity

Who cares about Richland? :p

Temash is faster than Brazos at 1/3rd TDP.

Actually the Z-60 has a TDP of 4.5W, so they are the same in that regard (TDP of less than 5W for Temash)

*hmm I think I misread your post, I guess you were referring to Temash vs. E2-1800. Along that line of thought it is also interesting to note that the A8-5545 has almost the same GPU performance as the A10-4600, so they've managed to take the GPU performance of trinity at a TDP of 35W and fit it into a TDP of just 19W with Richland (assuming the A8-5545 and A8-4555 has the same TDP), which is arguably even more impressive given that they are on the same process (32 nm)

[inf64]

Ok so Richland is 32nm after all with GCN(or not?)- clock bumps on CPU/GPU side? Good news is that Kaveri is H2 2013 product now and not 2014(unless things change of course).
Kabini/Temash look most impressive of all .

[podspi]

Say what you want, that presentation painted the rosiest picture of AMD's roadmap I've seen yet.

I still think 'connected standby' is a must have feature for those mobile SoCs though, still no word of that from the AMD side?

[Nemesis 1]

Originally Posted by Abwx

20% and 30% would be 25% average improvement , unless
you re using your usual Intel dedicated metric.....

50% average could mean 30% CPU and 70% GPU....

Watch the dang telacast she said it had total performance increase of 50 %The the cpu and the igpu didn't have 50% gains she said Total 50% improvement. I don't know how they arrived at that number . I do know thats not good enough tho . Its not worth debating Both chips will be arriving around the same time.

[dbcoopernz]

AMD has put some benchmark numbers for Kabini, Temash, and Richland on their website.

http://www.amd.com/us/press-releases..._new_apus.aspx

Testing and projections develop by AMD Performance Labs. The score for the 2012 AMD A10-4600M on 3DMark 11 was 1150 and the 2012 AMD A8-4555M was 780 while the “Richland” 2013 AMD A10-5750M was 1400 and the AMD A8-5545M was 1100. PC configuration based on the “Pumori” reference design with the 2012 AMD A10-4600M with Radeon™ HD 7660G graphics, the 2012 AMD A8-4555M with AMD Radeon™ HD 7600G graphics, the 2013 AMD A10-5750M with AMD Radeon™ HD 8650G graphics and the 2013 AMD A8-5545M with AMD Radeon™ 8510G Graphics. All configurations use 4G DDR3-1600 (Dual Channel) Memory and Windows 7 Home Premium 64-bit. RIN-1

Test conducted in AMD Performance Labs using FutureMark 3DMark Vantage P as a metric for GPU performance and PCMark Vantage as a metric for CPU performance. The AMD Z-60 APU-based system scored 455 and 1914 in 3DMark and PCMark respectively. The AMD A6-1450 APU-based system delivers scores of 981 and 3123 in 3DMark and PCMark respectively. The platforms tested were the AMD Z-60 APU with AMD Radeon™ HD 6250 graphics, 2x 2GB DDR3-1066, Microsoft Windows 7 and the “Larne” reference platform with an AMD A6-1450 Quad Core 1.0GHz APU, AMD Radeon™ HD 8280 Series graphics, 2GB DDR3-1066 system memory and Microsoft Windows 7. TEM-2

Test conducted in AMD Performance Labs measuring productivity performance with PCMark Vantage. The "Kabini" A6 APU-based system scored 5271 while the "Brazos" APU-based system scored 2807. “Kabini” PC config is based off the "Larne" reference design with 2013 AMD A6-5200 APU with AMD Radeon HD 8400 graphics, 4G DDR3 1600, and Windows 8 64bit. "Brazos" PC config is based off the "Renmore" reference resign with 2012 AMD E2-1800 APU with AMD Radeon HD 7340 graphics, 4G DDR3 1333 and Windows 7 Ultimate. KBN-3

Testing and projections develop by AMD Performance Labs. The score for the 2012 AMD A10-4600M on the PCMark 7 Overall benchmark was 1965 and the 2012 AMD A8-4555M was 1650, while the “Richland” 2013 AMD A10-5750M was 2175 and the 2013 AMD A8-5545M was 1850. PC configs based on the “Pumori” reference design with the 2012 AMD A10-4600M with Radeon™ HD 7660G graphics, the 2012 AMD A8-4555M with AMD Radeon™ HD 7600G graphics, the 2013 AMD A10-5750M with AMD Radeon™ HD 8650G graphics and the 2013 AMD A8-5545M with AMD Radeon™ 8510G Graphics. All configs used 4G DDR3-1600 (Dual Channel) Memory and Windows 7 Home Premium 64-bit. RIN-4

Based on testing and calculations by AMD Performance Labs, the 2013 AMD A-Series "Richland" platform enables 780 min./13:00 hours of Screen on idle / eReader. Battery life calculations based on a 6 cell Li-Ion 62.16Whr battery pack at 98% utilization. AMD defines "all-day" battery life as 8+ hours of battery life in Windows Idle. Test configuration: "Pumori" reference board, AMD A6-5350M APU with AMD Radeon™ HD 8450G Graphics, 4G DDR3-1600 (Dual Channel) Memory and Windows 7 Home Premium 64-bit, 14" 1366 x 768 eDP Panel / LED Backlight set at 100 nits, HDD (SATA) - 250GB 5400rpm, 62Whr Battery Pack and Windows 7 Home Premium. RIN-5

Some calculations:

The score for the 2012 AMD A10-4600M on 3DMark 11 was 1150 and the 2012 AMD A8-4555M was 780 while the “Richland” 2013 AMD A10-5750M was 1400 and the AMD A8-5545M was 1100.

- 22% and 41% in 3DMark 11 for A10 and A8, respectively. (Richland)

The AMD Z-60 APU-based system scored 455 and 1914 in 3DMark and PCMark respectively. The AMD A6-1450 APU-based system delivers scores of 981 and 3123 in 3DMark and PCMark respectively.

- 115% and 63% in 3DMark Vantage P and PCMark respectively. (Temash)

The "Kabini" A6 APU-based system scored 5271 while the "Brazos" APU-based system scored 2807.

- 88% in PCMark Vantage (Kabini)

The score for the 2012 AMD A10-4600M on the PCMark 7 Overall benchmark was 1965 and the 2012 AMD A8-4555M was 1650, while the “Richland” 2013 AMD A10-5750M was 2175 and the 2013 AMD A8-5545M was 1850.

- 11% and 12% in PCMark 7 for A10 and A8 respectively (Richland)

Based on testing and calculations by AMD Performance Labs, the 2013 AMD A-Series "Richland" platform enables 780 min./13:00 hours of Screen on idle / eReader.

(A6 Richland)

[Nemesis 1]

Originally Posted by podspi

Say what you want, that presentation painted the rosiest picture of AMD's roadmap I've seen yet.

I still think 'connected standby' is a must have feature for those mobile SoCs though, still no word of that from the AMD side?

I believe the highest Cstate is 6 . maybe 7 but 6 is what I recall . I thought the AMD pub cast would be better . But intels was better in my opion they demoed the gestures and face movements and eyemovement really well compared to what AMD showed . The Haswell at 7tdp was a pleasant surprize. I am more a Desktop guy and the direction this thing is going I not likeing much .

[Nemesis 1]

Originally Posted by dbcoopernz

The score for the 2012 AMD A10-4600M on 3DMark 11 was 1150 and the 2012 AMD A8-4555M was 780 while the “Richland” 2013 AMD A10-5750M was 1400 and the AMD A8-5545M was 1100.

- 22% and 41% in 3DMark 11 for A10 and A8, respectively. (Richland)

Thats some strange math your doing . the older one is the number thats used

You went 1400 is what % of 1150. Thats not right . The old processor ran at 1150 and the new one at 1400 Which came first the chiken or the egg .

I gotWhy do you want to use the new mark first . You want to go 1400 is what 18 % increase and 30 % without fussy mathCorrectly done it goes the old processor ran it at 1150 the new one at 1400 so it 1150 is what % of 1400= 82% 82-100= 18%

You want to write it 1400 is what % of 1150 =122% 122-100= 22 You are comparing the old to the new not the new to the old fussy math

[dbcoopernz]

Originally Posted by Nemesis 1

Thats some strange math your doing . the older one is the number thats used

You went 1400 is what % of 1150. Thats not right . The processor ran at 1150 and the new one at 1400 Which came first the chiken or the egg .

I got 18 % increase and 30 % without fussy math

I was comparing the Richland A10 to the Trinity A10. Why would I not state increase over Trinity as the ratio of Richland to Trinity.

[Homeles]

Originally Posted by dbcoopernz

I was comparing the Richland A10 to the Trinity A10. Why would I not state increase over Trinity as the ratio of Richland to Trinity.

Because you're not insane.

[piesquared]

Kitguru charted some extrapolations from performance claims by intel and AMD.



http://www.kitguru.net/components/gr...ance-analysed/

Richland would still have a healthy performance lead over haswell, provided the numbers are accurate. I assume AMD's numbers are since they show actual scores with their performance claims. Has intel given any performance numbers to back up it's 2x increase over ivy bridge? Couldn't find.

[Nemesis 1]

Because trinity was first . So you take the trinity number and get % of the Richland which tells you how much faster richland is over trinity. I understand you want to use the bigger % increase but I don't believe thats the correct method to find an increase over an older product . Your way is fine I will just remember to use the same method The new 5 over the old . thats sounds great . I think the old is how much slower is the old compared to the new you get 82% so you subtract 82 from 100 % to get how much faster the new is which is the new 100% 1150/1400 x y/100 = 82.14 so its 18%

[Nemesis 1]

You think it should look like this 1400/1150x y/100=122 so you get 22%. I don't care how its done . I just want to do it the same way down the road . If ya get my meaning .

[Nemesis 1]

Originally Posted by Homeles

Because you're not insane.

Thats not allowed in the cpu section . I don't care I just giving heads up .Pretty hard to insult me . pretty thick skinned

[dbcoopernz]

Yeah, if the benchmark is a "score" where higher is better, then new/old is the logical way to go. If the benchmark is a "time to completion" style (like Sunspider), then old/new is the way to go (e.g. 100ms for new vs 200ms for old gives a new part that is twice as fast).

[Nemesis 1]

Originally Posted by inf64

Ok so Richland is 32nm after all with GCN(or not?)- clock bumps on CPU/GPU side? Good news is that Kaveri is H2 2013 product now and not 2014(unless things change of course).
Kabini/Temash look most impressive of all .

It released in the second Qt that means no products till the end of 2013 or beginning of 2014 Right inline with intels silvermont release Which is 60% faster than the 32nm atom WITH HT.PER CORE

[Nemesis 1]

Originally Posted by dbcoopernz

Yeah, if the benchmark is a "score" where higher is better, then new/old is the logical way to go. If the benchmark is a "time to completion" style (like Sunspider), then old/new is the way to go (e.g. 100ms for new vs 200ms for old gives a new part that is twice as fast).

Thats OK . I don't care but for sandy we did the same thing and did it the other way . Numerator denuminator who new. expressed as a fraction than it would be 1400/1150 correct the numerator being 1400

[frozentundra123456]

Originally Posted by piesquared

Kitguru charted some extrapolations from performance claims by intel and AMD.



http://www.kitguru.net/components/gr...ance-analysed/

Richland would still have a healthy performance lead over haswell, provided the numbers are accurate. I assume AMD's numbers are since they show actual scores with their performance claims. Has intel given any performance numbers to back up it's 2x increase over ivy bridge? Couldn't find.

By the article's own admission, these are very speculative numbers based on each manufacturer's "best case" estimate. Lets not get too excited until we see real numbers by independent reviewers over a wide range of games.

Even using these projected best case numbers, one game is totally unplayable, and I would expect other demanding current and future games would be unplayable as well. Not to mention that by then the 8000 series cards will be out and hopefully will show a performance increase as well. So these igps will still trail low/mid range discrete cards by a huge amount. I just cant get excited about APUs in the desktop. Perhaps the lower power chips in tablets will be quite interesting though.

Edit: Based on the experience with my current Acer A100, I would never consider another Android tablet. I would only consider x86 running Windows, or (gasp) Apple.