I see no such result on the page you linked, so I guess they already removed the bogus results
Yes they're pretty quick in that. However it's still strange that something as flawed can actually pass to public display.
NVIDIA Tegra Architecture
- Thread starter french toast
- Start date
-
- Tags
- nvidia
Yes they're pretty quick in that. However it's still strange that something as flawed can actually pass to public display.
http://blogs.wsj.com/digits/2012/09/14/nvidia-confirms-departure-of-mobile-chip-chief/
It looks like Mike Rayfield has left NV.
It looks like Mike Rayfield has left NV.
Didn't Charlie @ SemiAccurate predict that Mike Rayfield was on his way out about 2 years ago?
It is not too surprising that Mike resigned. It's been nearly 3/4 of a year since Tegra 3 first debuted in a commercial device as the world's first mobile processor with quad-core CPU, and there is only one Tegra 3 smartphone design win (HTC One X international) listed on their website (http://www.nvidia.com/object/tegra-superphones.html). Of course, many more Tegra 3 smartphones are on the way (mainly from Chinese vendors, and some others such as LG Optimus 4X), but this is still not a great result. And we have yet to see even one Tegra 3 LTE device yet (although this is reportedly coming soon in the form of HTC One X+). NVIDIA is approximately 6 months behind Qualcomm with LTE and 28nm fabrication on mobile devices. The saving grace for Tegra 3 has been tablet design wins (although even that could be much better if Amazon, Barnes & Noble, and some others were on board), but the lack of smartphone design wins in 1H 2012 (especially with Motorola jumping ship) has surely been a disappointment for the company.
On a side note, while I'm sure Mike is a very competent and intelligent person, in interviews he never came across to me as particularly passionate or enthusiastic. In fact, his delivery style was pretty flat. NVIDIA needs a leader for their Tegra mobile group who is extremely personable, extremely passionate, extremely sharp, and extremely enthusiastic. NVIDIA's CEO and Chief Scientist are a couple examples of people who in interviews come across to me as very sharp, very intelligent, and very passionate. At the end of the day, there has to be some sense of urgency and accountability in executing well on the mobile front, as the competition is intense and unrelenting at the moment.
Interestingly enough, NVIDIA did recently hire Bob Feldstein from AMD. I would imagine that he would be assigned to mobile and console projects, but it is unclear who would be a proper fit for the head of the Tegra mobile group.
Tegra 4 (and by extension Tegra 5) will be of extreme importance for NVIDIA, because the expectation is that many mistakes or miscalculations (relating to hardware design, software design, roadmap execution, etc.) made in the past with Tegra 2/3 will be corrected. Based on what I've read, I expect the performance gap to be quite large vs. prior Tegra architectures, but the same can be said about the competition too.
Of course, next up is Tegra 3+ (T37). The expectation here is ~ 20-30% CPU and GPU performance improvement vs. Tegra 3. We know that the CPU operating frequency will be ~ 20-30% higher, but can we really expect a GPU operating frequency that is ~ 20-30% higher too considering that the ULP Geforce already has an operating frequency up to 520MHz?
It is not too surprising that Mike resigned. It's been nearly 3/4 of a year since Tegra 3 first debuted in a commercial device as the world's first mobile processor with quad-core CPU, and there is only one Tegra 3 smartphone design win (HTC One X international) listed on their website (http://www.nvidia.com/object/tegra-superphones.html). Of course, many more Tegra 3 smartphones are on the way (mainly from Chinese vendors, and some others such as LG Optimus 4X), but this is still not a great result. And we have yet to see even one Tegra 3 LTE device yet (although this is reportedly coming soon in the form of HTC One X+). NVIDIA is approximately 6 months behind Qualcomm with LTE and 28nm fabrication on mobile devices. The saving grace for Tegra 3 has been tablet design wins (although even that could be much better if Amazon, Barnes & Noble, and some others were on board), but the lack of smartphone design wins in 1H 2012 (especially with Motorola jumping ship) has surely been a disappointment for the company.
On a side note, while I'm sure Mike is a very competent and intelligent person, in interviews he never came across to me as particularly passionate or enthusiastic. In fact, his delivery style was pretty flat. NVIDIA needs a leader for their Tegra mobile group who is extremely personable, extremely passionate, extremely sharp, and extremely enthusiastic. NVIDIA's CEO and Chief Scientist are a couple examples of people who in interviews come across to me as very sharp, very intelligent, and very passionate. At the end of the day, there has to be some sense of urgency and accountability in executing well on the mobile front, as the competition is intense and unrelenting at the moment.
Interestingly enough, NVIDIA did recently hire Bob Feldstein from AMD. I would imagine that he would be assigned to mobile and console projects, but it is unclear who would be a proper fit for the head of the Tegra mobile group.
Tegra 4 (and by extension Tegra 5) will be of extreme importance for NVIDIA, because the expectation is that many mistakes or miscalculations (relating to hardware design, software design, roadmap execution, etc.) made in the past with Tegra 2/3 will be corrected. Based on what I've read, I expect the performance gap to be quite large vs. prior Tegra architectures, but the same can be said about the competition too.
Of course, next up is Tegra 3+ (T37). The expectation here is ~ 20-30% CPU and GPU performance improvement vs. Tegra 3. We know that the CPU operating frequency will be ~ 20-30% higher, but can we really expect a GPU operating frequency that is ~ 20-30% higher too considering that the ULP Geforce already has an operating frequency up to 520MHz?
Last edited by a moderator:
Considering the resources they've poured into up to now Tegra SoCs, it's still admirable that they ended up with as competitive offerings.
From the blog link above:
I don't think anyone would expect an IHV like NVIDIA to pour a huge amount of resources into a market they're fairly new and have to confront not one but quite a number of competing SoC manufacturers. It's a process that will build itself up step by step in the future.
3000 engineers and other employees. Nicely worded. Cleaning personell and Lord knows what else probably included in the 3k....
From the blog link above:
I don't think anyone would expect an IHV like NVIDIA to pour a huge amount of resources into a market they're fairly new and have to confront not one but quite a number of competing SoC manufacturers. It's a process that will build itself up step by step in the future.
3000 engineers and other employees. Nicely worded. Cleaning personell and Lord knows what else probably included in the 3k....
D
Deleted member 13524
Guest
Tegra 3 is everything but a disappointment to nVidia..
In the last financial results, the Tegra division was reported to carry huge profits and growth.
Take away the Galaxy Note 10.1 and every single high-end tablet has a Tegra 3.
I don't really think nVidia ever had high hopes for Tegra 3 in smartphones. It is, after all, still made on 45nm, unlike its quad-core competitors.
In the last financial results, the Tegra division was reported to carry huge profits and growth.
Take away the Galaxy Note 10.1 and every single high-end tablet has a Tegra 3.
I don't really think nVidia ever had high hopes for Tegra 3 in smartphones. It is, after all, still made on 45nm, unlike its quad-core competitors.
In the tablet space, Tegra 3 has certainly met and even exceeded NVIDIA's expectations, no doubt about it. Tegra revenue will continue to grow this year due to the upcoming launch of Tegra 3 Windows RT tablets, and due to the upcoming launch of affordable Tegra 3 smartphones. Starting end of this year or early next year, NVIDIA's outlook in both the smartphone space and tablet space will start to look even better. Their acquisition of Icera will finally start to yield dividends through use of a software defined radio baseband processor, which can be used on 3G/4G LTE smartphones and tablets. Their use of 28nm fabrication process will mean substantially lower power consumption for the same level of performance, or substantially higher performance for the same level of power consumption, vs. their current 40nm (not 45nm!) fabrication process, without the need to worry about 28nm supply constraints. Their use of multiple SKU's (aka "Batman", "Robin", etc.) will result in higher performance SoC's for high end tablets in addition to lower performance SoC's for lower end tablets and smartphones, meaning much better competitiveness vs. other high end tablets. Their use of Kepler DNA will result in much improved GPU performance per watt and GPU performance per mm^2. So the future is looking very rosy indeed, but clearly Mike was not a good fit for the company for whatever reason, and the performance of the Tegra mobile division from 2008 through 1H 2012 was not extraordinary.
Could it have been extraordinary under any of the circumstances and is it really fair after all to put the blame/responsibility on just one individual whoever that is? In such a big company where decisions/strategies don't fall on just indvidual but are rather a collective responsibility it's rather unfair IMHO to try to look for scape goats. I'm not aware of any finer inside details, but I'm sure that Rayfield was responsible for the lack of resources f.e. or the simple fact that NVIDIA posed way too optimistic in any of its own past projections.
Likewise it took Intel eons to make a half way decent kickstart in the smartphone market; do we have a name to use as an equivalent scape goat there or is it just that things aren't just as simple as some make them to be?
Unless of course you're working for the company and can't reveal details we regular "mortals" can't know; but I'd still severely doubt that it would change the overall reasoning above.
How about 20--30% more SoC bandwidth than in the T700 with the ULP GF clocked again at 520MHz?
Well life is not fair. If any group within a company is perceived to be underperforming, then the manager of that group is usually held accountable, and that is reasonable grounds for termination. Alternatively, if a manager is not a good leader and/or doesn't command the utmost admiration and respect of the group, then that is reasonable grounds for termination. As another alternative, the company may have been willing to reassign him to a different role, but he preferred to resign rather than being demoted. And as yet another alternative, sometimes people resign for purely personal reasons or because they want to try something new (which is very common when people work at a company for 5+ years). This is all purely speculative, and you and I certainly have no way of knowing the truth without having the luxury of talking to Mike or an insider at the company. Looking at previous interviews online, he seemed very laid back and lacked emotion and conviction in his speech (signifying that he may not have been a good fit with more aggressive types of people at the company), but that could easily be overanalysis on my part. Anyway, Charlie could have been right in his commentary a while back about upper management not being satisfied with the performance and roadmap of the Tegra mobile division, but like always I'm sure the truth of the matter behind the recent resignation is much more complicated than that. The puzzling part is, if Charlie heard these rumors in the middle of 2010, then why a sudden and quiet resignation 2 full years later? Who knows for sure, but maybe Charlie can dig up something new . The only thing I can think of is that there has been some serious ongoing tension between Mike and other key member(s) of the Tegra mobile group over the last few years, and this may have contributed to Mike's resignation, but again this is all purely speculative.
. The only thing I can think of is that there has been some serious ongoing tension between Mike and other key member(s) of the Tegra mobile group over the last few years, and this may have contributed to Mike's resignation, but again this is all purely speculative.
Last edited by a moderator:
Well life is not fair. If any group within a company is perceived to be underperforming, then the manager of that group is usually held accountable, and that is reasonable grounds for termination. Alternatively, if a manager is not a good leader and/or doesn't command the utmost admiration and respect of the group, then that is reasonable grounds for termination. As another alternative, the company may have been willing to reassign him to a different role, but he preferred to resign rather than being demoted. And as yet another alternative, sometimes people resign for purely personal reasons or because they want to try something new (which is very common when people work at a company for 5+ years). This is all purely speculative, and you and I certainly have no way of knowing the truth without having the luxury of talking to Mike or an insider at the company. Looking at previous interviews online, he seemed very laid back and lacked emotion and conviction in his speech (signifying that he may not have been a good fit with more aggressive types of people at the company), but that could easily be overanalysis on my part. Anyway, Charlie could have been right in his commentary a while back about upper management not being satisfied with the performance and roadmap of the Tegra mobile division, but like always I'm sure the truth of the matter behind the recent resignation is much more complicated than that. The puzzling part is, if Charlie heard these rumors in the middle of 2010, then why a sudden and quiet resignation 2 full years later? Who knows for sure, but maybe Charlie can dig up something new
If the Tegra department was lacking in resources (which I haven't read or heard just once up to now) then it wasn't obviously just Rayfield's responsibility. There's a note above in the link that states that given the resources he had for his department its performance was fine.
Yes there might have been disagreements between Rayfield within his own department or even if the upper management of the company which might have been in parts related to a possible lack of resources.
As I said it will be highly interesting to find out where Rayfield will work in the future as it might show that he simply got a way better offer or might have found a job with higher ambitions elsewhere or maybe even not. If it should be the first case and the department he might be leading shows a sudden success rate it'll be Rayfield's and any Rayfield's partial success since large departments like the Tegra department aren't exactly a one man show where failure or success isn't just one individual's success.
Maybe Charlie would like to write an article revealing how the resource story exactly went for the Tegra department.
Tweet from Paul O'Brien: "Yowza. The One X+ seriously rocks the benchmarks! Early software, over 7500 in Quadrant, nearly 14k in Antutu, over 150 MFLOPS in Linpack!".
Considering that the HTC One X+ is supposed to have Jellybean OS installed, and considering that the One X+ is supposed to be at least ~ 20-30% faster than the One X, then the numbers make sense. Now, the Linpack number must be the Multi-Threaded benchmark, so nothing special about that, especially compared to Krait.
Considering that the HTC One X+ is supposed to have Jellybean OS installed, and considering that the One X+ is supposed to be at least ~ 20-30% faster than the One X, then the numbers make sense. Now, the Linpack number must be the Multi-Threaded benchmark, so nothing special about that, especially compared to Krait.
One X+ officially announced
http://www.anandtech.com/show/6348/...official-17-ghz-ap37-tegra-3-64-gb-of-storage
Anandtech claims its still on 40nm though
http://www.anandtech.com/show/6348/...official-17-ghz-ap37-tegra-3-64-gb-of-storage
Anandtech claims its still on 40nm though
Nebuchadnezzar
Legend
The normal One X is already at roughly half the battery life of the S3, so they throw in a bigger battery, but compensate that with a even more power hungry SoC? No wonder they quote just talk time and ignore screen time. This may be still acceptable in the tablet space with much larger batteries, but they're shooting themselves in the foot in the mobile space by still being on 40nm.anexanhume
Veteran
Nvidia is in a weird place with 40nm right now. They're essentially on the same release schedule as TI with OMAP 5, and everyone knows the trouble TI is going through. Qualcomm and Apple caught the rest of the industry at a bad time as they do reference designs while they can do custom ones and get them out quicker.
Battery life should not be a concern with the One X+ for a variety of reasons: HTC is using a significantly larger battery (as you noted); the AP37 Tegra 3 variant used in the One X+ should be a lower leakage part vs. the AP33 Tegra 3 variant used in the One X; updated software by HTC/NVIDIA has reportedly improved battery life significantly since the One X was introduced 6 months ago; there is a new battery saver mode which reduces the CPU operating frequencies for those who want to enhance battery life during heavy use.
The One X+ is actually pretty darn impressive all things considered. The Quadrant, Antutu, and presumably Geekbench scores are as high or higher than virtually any other smartphone on the market today, including the quad core Krait S4 Pro equipped LG Optimus G! Of course, the Optimus G does have a faster GPU, but one could argue that the difference in GPU performance between these two smartphones is largely academic at this point, as the Tegra 3 equipped phones will have no problems running most DirectX 9 style games smoothly (such as those using Unreal Engine 3), and as Tegra 3 equipped phones will have access to some additional visual effects with TegraZone games.
Since Tegra 3+ appears to be very competitive overall compared to what else is available today, NVIDIA is actually in a great position leading up to Tegra 4, where the expectation is that we will see significant enhancements with Cortex A15-class CPU, Kepler-class GPU, Icera 3G/4G LTE baseband processor, 28 nm fabrication process, etc. Since new Tegra 3+ variants are being released now or very soon, I wouldn't expect to see Tegra 4 in phones for at least 6 months.
Similar threads
