I see one possibly flaw in your calculation. The A8-3520M in your link is able to score ~3000 points in Vantage. It means the 17W score is about 20-25% slower based on Vantage compared to AMDs predecessor 35W APU. So I'm not that sure if Trinity 17W is able to outscore the HD4000@1150 Mhz, especially not with a 25-40% gap because 1250 Mhz HD4000 vs A8-3520M are pretty close overall.
Ivy Bridge GPU details
- Thread starter dkanter
- Start date
I see one possibly flaw in your calculation. The A8-3520M in your link is able to score ~3000 points in Vantage. It means the 17W score is about 20-25% slower based on Vantage compared to AMDs predecessor 35W APU. So I'm not that sure if Trinity 17W is able to outscore the HD4000@1150 Mhz, especially not with a 25-40% gap because 1250 Mhz HD4000 vs A8-3520M are pretty close overall.
With turbo yes, but how often will that be available on a 17W chip is the whole point surely? From the article on RWT intel says the mobile IGP will not exceed 12W - clearly there is not much chance of it getting anywhere near that on a 17W TDP chip.
As I mentioned in my previous post there is a large difference (33%) in Vantage scores between the 45W and 35W Ivy's. That's with only a 29% reduction in TDP - the 17W will have half the TDP. I just can't see it scoring much higher than 2k in Vantage tbh which surely means it will be quite far behind in actual gaming benchmarks compared to the 17W Trinity.
The 17W Ivy will practically never reach 1150 MHz. It's little more than a marketing number, the true speed will barely go above 500 MHz I feel.
As for Trinity, at 17W bandwidth will be less of a contraint on the graphics. Even those 35W Llano's are bandwidth constrained but obviously this become less of an issue with less powerful chips. That 25% slower than the 35W Llano is about where I expect the 17W Trinity to be in gaming.
I doubt it. With just 500 Mhz it would be slower than HD3000 ULV. This slide implies a GPU Turbo for Trinity mobile as well.
It won't be slower because it has more EU's, and we don't know what the "average" clock speed of SB ULV's are anyway.
Bit of a leak here - http://www.cpu-world.com/news_2012/2012051001_AMD_readies_R-series_embedded_APUs.html
Bit of a leak here - http://www.cpu-world.com/news_2012/2012051001_AMD_readies_R-series_embedded_APUs.html
The difference isn't that big with HD 3000. 2600K can get 2100, and 35W dual core mobile parts get ~1900. 17W chip does 1600.
We've got better comparisons. The A6-4455M(which we know it to be a 17W part) gets 620 in 3DMark11 using DDR3-1600. 3DMark11 seems to be a pretty accurate benchmark when comparing the most recent iGPUs.
Now the unknown is Ivy Bridge. If they can pull off what they did with HD 3000, then it might be on par with Trinity 17W. Of course the HD 3000 had big differences depending on vendors. Series 9 laptop for example only performed at 2/3rds of other 17W ones. Tier 1 vendors like Asus can get within 15-20% of top non-17W performance.
The Zenbook is languishing some 115% slower than the a8-3500M in metro 2033 -
http://www.anandtech.com/show/4985/asus-zenbook-ux21-review/6By my reckoning the 17W Trinity will score ~ 20fps here. It's asking a lot for the 17W Ivy to hit 20 fps I feel, 16-17 fps is more likely.
It'll be pretty close in some games but over the piece I'll be pretty amazed if Trinity isn't ahead in most of them. AMD is saying 18-19% on their "projections" I believe.
http://blogs.amd.com/play/2012/05/1...ring-your-ultrathin-and-light-gaming-machine/ It's actually hard to figure out what he's saying there because of not naming the intel cpu's though.
Looking at the previous link from cpu world AMD appears to be saying they are 50% faster at 35W (even vs intel's 45W). I just can't see them losing all of that at 17W.
Last edited by a moderator:
115% slower? Negative speed? It runs backwards in time?
Or do I see that right, that you define "slowness" as the inverse of "speed"?
So you set the A8-3500M as base at 100% slowness, which makes the Zenbook 115% slower and therefore having a slowness value of 215%. The speed is the inverse of it: 46.5%. Or the A8 is 115% faster.
SCNR.
Last edited by a moderator:
Yeah, seriously let's stop this confusing "%" BS already. This is Beyond3D guys, everyone understands simple fractions.115% slower? Negative speed? It runs backwards in time?
115% slower? Negative speed? It runs backwards in time?
Or do I see that right, that you define "slowness" as the inverse of "speed"?
So you set the A8-3500M as base at 100% slowness, which makes the Zenbook 115% slower and therefore having a slowness value of 215%. The speed is the inverse of it: 46.5%. Or the A8 is 115% faster.
SCNR.
Yep I realised it after I wrote it, then I tried to change a few things but the whole post went crazy with a paragraph link to anandtech which I couldn't edit. After that I really couldn't be arsed and figured nobody here could surely be pendantic enough to pull me up for it!
4600m benchmarks - that's a huge increase in graphics performance in Vantage but not so much with 11 (page 3).
http://translate.google.com/translate?sl=zh-CN&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&u=http%3A%2F%2Fnotebook.pconline.com.cn%2Ftesting%2Fcontrast%2F1205%2F2781947.html&act=url
http://translate.google.com/translate?sl=zh-CN&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&u=http%3A%2F%2Fnotebook.pconline.com.cn%2Ftesting%2Fcontrast%2F1205%2F2781947.html&act=url
OpenGL guy
Veteran
Much better to understand irrational numbers as there are a lot more of them
Lol, very cleverMuch better to understand irrational numbers as there are a lot more of them
Similar threads
