Nvidia jumped from 40nm to 28nm and cortex A9 to A15 so this results into bump. While Qualcomm was already using latest 28nm technology and Krait was very close to Cortex A15. So we can call this performance increase huge.
"And the bump in Tegra 4's GPU perf is independent of the process transition. They seem to have bumped the number of GPU cores to 6x in Tegra 3."
What? Its directly related. By moving to 28nm nV was able to fit A15 AND more shaders in the same die area as their 40nm process. They couldn't have used that many shaders without ballooning die size and TDP. So actually, the bump in Tegra 4's GPU perf is hugely dependent on the process transition
"Qualcomm tells us that the Adreno 330 will offer roughly 50% more graphics performance over Adreno 220, and an almost 2x increase in compute performance."
@pugster: unlike NVIDIA, Qualcomm's Adreno is one of the leading mobile GPUs out there, they don't need to play catch-up like NVIDIA does. NVIDIA's GeForce ULPs found in Tegra 2 and 3 were rather lacking for a while. Qualcomm can keep the 800 series very competitive by just improving graphics by 50% to compete with NVIDIA.
till now only Adreno was able to challenge Apple Soc but this year Mali 658, Nvidia GeForce and Adreno 330 all Challenge Apple Soc. So this will be great.
Well, in terms of graphics performance(!) the announced SoCs are a bit disappointing. 50% more graphics performance over Adreno 320 doesn't even beat last years Apple A6X. And the Tegra 4's graphics performance is probably also not higher than the A6X's, or Nvidia would have said so it in their presentation (not a single graphics performance comparison...). Unless Rogue is a huge disappointment, Apple has nothing to worry about in 2013, at least from Nvidia and Qualcomm.
Does it matter? If you want an iPhone, you're going to get an iPhone. No one cross-shops iOS and Android based on SoC performance. That's exclusively an Android thing (and even then, you pick phones, not SoCs).
I swear, they just implemented this new "S" architecture (which was kind of stupid) and now they're changing the branding scheme again? Jeez, the only people who actually care about the brand/model name of their SoC still refer to them by the MSM/APQXXXX naming scheme, so they should just cut the crap
The S naming made sense, as well as any model numbering system can.
S3 was the Scorpion CPU-based SoC. S4 was the original Krait CPU-based SoC. And having multiple levels in each made sense as a way to differentiate performance levels (S4 Play, S4 Pro, etc). They should have continued with it, and these new SoCs would be S5 (for Krait 300-based SoCs) and S6 (for Krait 400-based SoCs).
That would have made sense, and provided continuity across at least 3 years of SoCs.
Switching from S1-S4 to S200-S800 is just plain old marketing "biggar numbars whar!". :(
Unfortunately, Qualcomm has a deep hatred for anything resembling logical part numbering. Gotta love how you can have Scorpion and Krait-based SoCs with the same MSM8xxx part number. Or how lower MSMxxxx numbers can be newer than higher MSMxxxx numbers.
Maybe, eventually, they'll figure this all out. But I'm not holding my breath waiting.
They introduced "S" numbering system because they incorporated many older SoCs too in the main fold and hence wanted people to know which one is an older generation. (like S1 is oldest and S4 is newest). And they have changed to this strategy now because it allows them to classify newer chips based on the performance and not on the year they were made available to customers. This will also allow them flexibility to use mid-way numbers like 250, 300, 350 etc. which may aim at both noting the performance boost and also the launch year.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
22 Comments
Back to Article
pugster - Monday, January 7, 2013 - link
Not much bump of GPU power compared to their precessor, unlike Nvidia's Tegra4akmittal - Tuesday, January 8, 2013 - link
Nvidia jumped from 40nm to 28nm and cortex A9 to A15 so this results into bump. While Qualcomm was already using latest 28nm technology and Krait was very close to Cortex A15. So we can call this performance increase huge.varad - Tuesday, January 8, 2013 - link
"Krait was very close to Cortex A15":You might want to back that up. Here are some numbers measured by Anandtech:
http://www.anandtech.com/show/6425/google-nexus-4-...
And the bump in Tegra 4's GPU perf is independent of the process transition. They seem to have bumped the number of GPU cores to 6x in Tegra 3.
akmittal - Tuesday, January 8, 2013 - link
from close to A15 ,i meant much better than Cortex A9Activate: AMD - Tuesday, January 8, 2013 - link
"And the bump in Tegra 4's GPU perf is independent of the process transition. They seem to have bumped the number of GPU cores to 6x in Tegra 3."What? Its directly related. By moving to 28nm nV was able to fit A15 AND more shaders in the same die area as their 40nm process. They couldn't have used that many shaders without ballooning die size and TDP. So actually, the bump in Tegra 4's GPU perf is hugely dependent on the process transition
Assimilator87 - Monday, January 7, 2013 - link
"Qualcomm tells us that the Adreno 330 will offer roughly 50% more graphics performance over Adreno 220, and an almost 2x increase in compute performance."Did you mean "over Adreno 320"?
akmittal - Tuesday, January 8, 2013 - link
Yup they mean Adreno 320 not 220..Brian Klug - Tuesday, January 8, 2013 - link
Oops that's right, fixed!-Brian
deltatux - Monday, January 7, 2013 - link
@pugster: unlike NVIDIA, Qualcomm's Adreno is one of the leading mobile GPUs out there, they don't need to play catch-up like NVIDIA does. NVIDIA's GeForce ULPs found in Tegra 2 and 3 were rather lacking for a while. Qualcomm can keep the 800 series very competitive by just improving graphics by 50% to compete with NVIDIA.mayankleoboy1 - Tuesday, January 8, 2013 - link
With the Adreno330, can they beat PowerVr in Apple SoC's ?2013 is definitely going to be exciting for ARM SoC's . We have Exynos 5xxx, T4 and Krait 300.
And of course, Apple.
akmittal - Tuesday, January 8, 2013 - link
till now only Adreno was able to challenge Apple Soc but this year Mali 658, Nvidia GeForce and Adreno 330 all Challenge Apple Soc. So this will be great.pyaganti - Tuesday, January 8, 2013 - link
The current Exynos5xxx is using T604 right? Are they releasing Quadcore Exynos with T658?Mike1111 - Tuesday, January 8, 2013 - link
Well, in terms of graphics performance(!) the announced SoCs are a bit disappointing. 50% more graphics performance over Adreno 320 doesn't even beat last years Apple A6X. And the Tegra 4's graphics performance is probably also not higher than the A6X's, or Nvidia would have said so it in their presentation (not a single graphics performance comparison...). Unless Rogue is a huge disappointment, Apple has nothing to worry about in 2013, at least from Nvidia and Qualcomm.jeffkibuule - Tuesday, January 8, 2013 - link
Does it matter? If you want an iPhone, you're going to get an iPhone. No one cross-shops iOS and Android based on SoC performance. That's exclusively an Android thing (and even then, you pick phones, not SoCs).Chloiber - Tuesday, January 8, 2013 - link
When will they be released?akmittal - Tuesday, January 8, 2013 - link
Q2 this year, likely with Galaxy S4amdwilliam1985 - Tuesday, January 8, 2013 - link
Nope, Galaxy S4 will be made with Exynos 5 with and without LTE.karim128 - Tuesday, January 8, 2013 - link
"Also being announced today is the Snapdragon 600. This part integrates two Krait 300 cores running at up to 1.9GHz."According to their website, it actually has 4 Krait 300 cores, which BTW makes it even more impressive for a mainstream offering !
http://www.qualcomm.com/media/blog/2013/01/07/snap...
Activate: AMD - Tuesday, January 8, 2013 - link
I swear, they just implemented this new "S" architecture (which was kind of stupid) and now they're changing the branding scheme again? Jeez, the only people who actually care about the brand/model name of their SoC still refer to them by the MSM/APQXXXX naming scheme, so they should just cut the crapphoenix_rizzen - Tuesday, January 8, 2013 - link
The S naming made sense, as well as any model numbering system can.S3 was the Scorpion CPU-based SoC. S4 was the original Krait CPU-based SoC. And having multiple levels in each made sense as a way to differentiate performance levels (S4 Play, S4 Pro, etc). They should have continued with it, and these new SoCs would be S5 (for Krait 300-based SoCs) and S6 (for Krait 400-based SoCs).
That would have made sense, and provided continuity across at least 3 years of SoCs.
Switching from S1-S4 to S200-S800 is just plain old marketing "biggar numbars whar!". :(
Unfortunately, Qualcomm has a deep hatred for anything resembling logical part numbering. Gotta love how you can have Scorpion and Krait-based SoCs with the same MSM8xxx part number. Or how lower MSMxxxx numbers can be newer than higher MSMxxxx numbers.
Maybe, eventually, they'll figure this all out. But I'm not holding my breath waiting.
The_Rex - Wednesday, January 9, 2013 - link
They introduced "S" numbering system because they incorporated many older SoCs too in the main fold and hence wanted people to know which one is an older generation. (like S1 is oldest and S4 is newest). And they have changed to this strategy now because it allows them to classify newer chips based on the performance and not on the year they were made available to customers. This will also allow them flexibility to use mid-way numbers like 250, 300, 350 etc. which may aim at both noting the performance boost and also the launch year.iwod - Wednesday, January 9, 2013 - link
Those improvement numbers sounds like Swift level of performance. Wondering which one will be faster clock per clock.