Not true. Not all the originally announced GS2 phones had Exynos processors, particular those going to Russia, Eastern Europe and other parts of the developing world. This tells me that Samsung's fabs are much smaller than their SoC rivals, not because of demand. Either that or the yields are low. If it was demand then they (most likely) would've either scaled production to meet demand or they would be announcing the building of bigger fabs.
Not true. Not all the originally announced GS2 phones had Exynos processors, particular those going to Russia, Eastern Europe and other parts of the developing world. This tells me that Samsung's fabs are much smaller than their SoC rivals, not because of demand. Either that or the yields are low. If it was demand then they (most likely) would've either scaled production to meet demand or they would be announcing the building of bigger fabs.
None of Samsungs competitors in the SoC market even have fabs. Samsung doesn't seem to have a problem with capacity as they still manufacture the SoC's used in every iPhone and iPad sold.
Last edited:
Fire&Blood
Platinum Member
- Jan 13, 2009
- 2,331
- 16
- 81
The first Tegra was a failure. 2nd was an average and the 3rd one appears to follow it's footsteps. Qualcomm's S3 chip is actually A8 Cortex based (custom Qualcomm core) yet it looks very good in direct comparison to Tegra2 and it's "stock" A9 Cortex core. Tegra3 can sell well through different tablet offerings but I don't see it making a dent in smartphone market share.
It's an aging, power hungry design. To accommodate for 5 cores, the chip lost a lot in raw transistor count. Since they had an inferior product on their hands, Nvidia was correct with the marketing "1st quad core" approach however they have to get past OEM's first to sell it that way. Trying to sell early before their chip's inferiority is exposed by established players. 28/32nm SoC's with A15 and custom variations will crush the Tegra and they know it, they are attempting to make a buck before they do.
Tegra lineup never yielded superior solutions no matter how hard Nvidia tries to market as such. One look at the specs reveals imposters:
It's an aging, power hungry design. To accommodate for 5 cores, the chip lost a lot in raw transistor count. Since they had an inferior product on their hands, Nvidia was correct with the marketing "1st quad core" approach however they have to get past OEM's first to sell it that way. Trying to sell early before their chip's inferiority is exposed by established players. 28/32nm SoC's with A15 and custom variations will crush the Tegra and they know it, they are attempting to make a buck before they do.
Tegra lineup never yielded superior solutions no matter how hard Nvidia tries to market as such. One look at the specs reveals imposters:
Not true. You are confused
The phones you are talking about had the model number I9103 and were sold under the name Samsung Galaxy R. At no point did Samsung sell them as "Samsung Galaxy S2". Eldar Murtazins russian GS2 was using Exynos and he was one of the first to review one.
So i would like to see some links to a russian Samsung Galaxy S2 without an Exynos in it considering im only a few hours away from Russia
And im not sure what you are talking about regarding Samsungs fab capability. Not only do they make a gazillion different SOCs for themselves, they made A4-A5 for Apple and the SOC for Playstation Vita
Wich one of Samsungs "SOC rivals" have their own fab let alone producing more?
The first Tegra was a failure. 2nd was an average and the 3rd one appears to follow it's footsteps. Qualcomm's S3 chip is actually A8 Cortex based (custom Qualcomm core) yet it looks very good in direct comparison to Tegra2 and it's "stock" A9 Cortex core. Tegra3 can sell well through different tablet offerings but I don't see it making a dent in smartphone market share.
It's an aging, power hungry design. To accommodate for 5 cores, the chip lost a lot in raw transistor count. Since they had an inferior product on their hands, Nvidia was correct with the marketing "1st quad core" approach however they have to get past OEM's first to sell it that way. Trying to sell early before their chip's inferiority is exposed by established players. 28/32nm SoC's with A15 and custom variations will crush the Tegra and they know it, they are attempting to make a buck before they do.
Tegra lineup never yielded superior solutions no matter how hard Nvidia tries to market as such. One look at the specs reveals imposters:
One thing that came into my mind was that this upcoming quarter will be the best quarter Nvidia could possibly ever get but they will get ruined for the rest of the year unless they gain some massive design wins like Kindle Fire
They were a lead licensee for Cortex A-9 wich allowed them to be first with dual-core A9. However they do not hold a lead license for Cortex A-15 wich means they will be last to market with that SOC.
This tells me that its highly unlikely that we will see a Tegra 4 in 2012 using Cortex A-15s. If it comes out next year it will be a Cortex A-9 of some sort
So while they were first with dual-core and first with quad-core, they will have to play catch up after that since everyone else will be on A-15 dual/quads with some seriously insane GPUs (MALI-T658/Power VR 544/Rogue)
Looking at the Tegra 3 benchmarks Anand has posted, it's hard to imagine it living up to that bad @ss press release. Press releases stomp all over actual products, as everyone knows.....
I guess the main thrust of this thread is that six months from now a new product will ship that is better then right now. I'm going to go utterly Nostradamus here, it's probably going to shock people- Ten years from now, everything we are talking about will look like trash next to an entry level Boost mobile phone. Bank on it, hold me accountable. This is a very difficult thing to see, something faster coming out later in the tech industry, I know, nothing like it has EVER happened before, but mark my words.
I guess the main thrust of this thread is that six months from now a new product will ship that is better then right now. I'm going to go utterly Nostradamus here, it's probably going to shock people- Ten years from now, everything we are talking about will look like trash next to an entry level Boost mobile phone. Bank on it, hold me accountable. This is a very difficult thing to see, something faster coming out later in the tech industry, I know, nothing like it has EVER happened before, but mark my words.
Red Storm
Lifer
- Oct 2, 2005
- 14,233
- 234
- 106
I don't particularly care about the CPU/GPU in a tablet, as long as it plays my media files then I'm good. I'm more interested in tangible things like screen quality and resolution. By all accounts, the Transformer Prime has a fantastic screen, but the resolution has me a little wary since we could see 1080P and higher res tablet displays in 2012.
Lenovo is releasing a Tegra 3 tablet with 2 gig ram and 1920x1200 resolution early next year. Might want to wait for that one atleast
Not sure how good Lenovo is handling updates though but if you are planning on running custom roms then i guess it doesnt matter
- Jun 23, 2001
- 27,730
- 8
- 0
Not sure if any of the people ranting about the inferiority of Nvidia's Tegra chips brought into account their tiny die sizes. Tegra 3 is ~80mm2, if I recall, at 40nm. This makes it far smaller than other SoCs. And there's always the potential for a 28nm die shrink without making any major architectural changes.
It is shocking information isnt it? I will predict 6 months after those chips are released or about a year from now there will be a good possibility Wayne will be faster than them.
Stop arguing for the sake of arguing, you clearly haven't done anything real world involving anything we're talking about. Take a single core processor and do something that consumes the majority of it's processing time and then try and run any non ancient game and you will have severe performance penalties in one or the other or both applications. If you do the same on a dual core there are no such penalties assuming each app was using only one thread. This has nothing to do with architecture what so ever; it does not matter how fast your single core processor is, this will always apply.
http://www.anandtech.com/show/1676/13
You're putting words in my mouth, I suggest you read the thread again. I said what I would prefer and I explained quite clearly why. I didn't say Tegra 3 is better than the new Exynos.
Tegra 3 has 100% more cores than this new Exynos, not 50% more so the Phenom II to SB comparison is invalid. Comparing an 8 core Bulldozer to a 4 core Sandybridge is not valid either because each core on Bulldozer takes a pretty large step back in performance compared to what was out while Tegra 3's 4 cores are state of the art as of now. Single thread performance in order of best to worst goes something like i7(SB), i7(Nehalem), Core 2 Penryn, Core 2 Conroe, Phenom II, Bulldozer.
A better comparison would be an 8 core Nehalem vs a 4 core Sandybridge.
Last edited:
Your prediction would be wrong. You are not looking at just another SOC, its a gigantic leap in technology. And im not talking about Samsungs version, im talking Cortex A-15 in general
Wayne is most likely another Cortex-A9 built on 28nm with a beefier GPU
Unfortunatly its going up against GPUS completely different than we have seen on mobiles so far. Rogue 272+ GFLOPs, MALI-T658 350 GFLOPS
Wayne is supposedly 10x more powerful than Tegra 2 while Tegra 3 was 5x more powerful than T2 wich makes Wayne 2x more powerful than T3 so all you need to do is calculate and understand that Wayne is DOA on arrival compared to what i mentioned above
Wayne is supposed to be a quad core SoC using A-15s. And doubling up the GPU core count to 24 from Kal-El. Nvidia managed to increase GPU performance by double going from 8 to 12 cores. I'd expect more when actually doubling them in Wayne.
- Jun 23, 2001
- 27,730
- 8
- 0
Shrug. Nvidia's not simply going to roll over and yield tom TI, Qualcomm, and Samsung. Current road maps can always be adjusted as products are re-aligned.
Everyone's neglecting the die size issue, and the claim that Tegra 3 will be found in sub-300 tablets within the year. Meaning that when the 3rd generation iPad launches, it'll be competing against a slew of inexpensive, relatively high performance, long battery life Tegra 3 tablets running Ice Cream Sandwich and (possibly)Windows 8. Competition and choice are a good thing.
They seem to be so far. Tegra 2 barely beat Exynos to the phone market yet Exynos wipes the floor with it.
It isn't so easy to realign roadmaps when you are dependent on outside forces such as ARM and TSMC. It was stated earlier that Nvidia did not have an early license for Cortex A15 which will surely put them at a disadvantage vs other companies that do.
Have to apologize as you could be correct about the Cortex A-15. I thought for some reason that Nvidia had skipped a lead license this round
Tegra 3 GPU supposedly runs at 400 mhz. Wich means it has roughly 9 GFLOPS (based on Anands preview)
Even if we quadruple that its still roughly just under 40 GFLOPS
the Nova SOC from ST-Ericsson is quoting Rogue does 272 GFLOPS
Considering an already aged SGX 543 MP2 gets 19 GFLOPS at 300 MHZ and could most likely reach 40 just by going to 28nm and overclocking to 500 mhz it makes Tegra 3 look even less impressive and i doubt Wayne will change that
Nvidias biggest strength is their software/drivers
Here you go:
http://www.engadget.com/2011/07/01/samsung-galaxy-z-galaxy-s-iis-affordable-little-brother-now/
http://www.engadget.com/2011/03/20/samsung-galaxy-s-ii-mini-leaks-out-for-three-uk-plus-nokia-x7/
http://www.engadget.com/2011/03/03/confirmed-tegra-2-equipped-samsung-galaxy-s-ii-is-coming/
http://www.engadget.com/2011/02/17/samsungs-galaxy-s-ii-to-have-a-tegra-2-version/
ATI's cards have double the theoretical processing bandwidth compared to Fermi based devices. In the real world when numbers are thrown at them. They typically get their doors blown off.
272 GLOPS. That would put it at roughly 1/4th the power of a GTX 560 Ti. You think that will be what we see in the real world? I will remain skeptical until we see that GPU pushing about 20 times the frames of the benchmarks we saw today for Tegra 3. While not consuming that battery in under a minute.
Thank you. So you are basically agreeing with me that the international Galaxy S2 contained Exynos even in Russia aswell as other regions. These phones were sold as "Samsung Galaxy Z, Samsung Galaxy R, Samsung Galaxy Mini" as cheaper alternatives to the flagship model
Fire&Blood
Platinum Member
- Jan 13, 2009
- 2,331
- 16
- 81
Nvidia should secure the Kindle Fire deal and other tablets as well with the Tegra3 as an early "almost next gen" processor compared to established players.
Tegra clearly doesn't belong in high end products, performance wise that's Exynos territory.
People need to get past impressive tech demos and staged benchmarks to realize Nvidia's mobile offerings are only average and that's what they are worth. If they are buying while aware, that's fine but it's an illusion to go with Tegra products and think you are buying high end.
I don't think that anyone is arguing that there is something better around the corner, it's more about the lack of value with Nvidia. If the bigger Kindle Fire with Tegra3 costs $300, why pay more for the Prime? It's not just that future chips will own Tegra, that's a given, it's that unlike the Exynos, Tegra never took the performance crown and doesn't hold value very well.
Tegra clearly doesn't belong in high end products, performance wise that's Exynos territory.
People need to get past impressive tech demos and staged benchmarks to realize Nvidia's mobile offerings are only average and that's what they are worth. If they are buying while aware, that's fine but it's an illusion to go with Tegra products and think you are buying high end.
That would hold water if it were not for the fact that the SGX 543MP2 released a year ago is already slapping Tegra 3 around in every single GPU benchmark available
And its underclocked at 250 mhz while Nvidias Tegra 3 GPU runs at 400 mhz. And still easily beating it
What wont hold water? I am doubting we will see a hand held or tablet GPU that will deliver 272GLOPS in 2012.
Nope, not agreeing with you. I wanted to show you that there were GSII with non-Exynos SoCs. I don't know or care who that Russian guy is but I do remember Samsung reserving the Exynos GSIIs for Western countries and giving others the Tegra-based GSIIs. Whoever is making SoCs for Samsung's rivals must be doing a better job then because their yields or fabs are bigger.
I meant your theoretical vs real world differences comparison. In real world performance a last gen SOC from IMG is already soundly beating Nvidias just released SOC
Rogue does not need to deliver 272 GFLOPS in order to beat Wayne. I mentioned it to give an idea of just how big of a leap this will be in architecture/technology.
Then you failed. Because you did not show a single Galaxy S2 running Tegra 2
http://www.samsung.com/ru/consumer/m...ail&returnurl=
Thats the same model number you tried to claim was a Galaxy S2. Yet Samsungs own russian website is calling it Samsung Galaxy R. If you go to the other phones at their russian site, you can find the real Galaxy S2 using Exynos and SAMOLED screen
Its irrelevant what you remember. You have posted zero factual information to back your claim up.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 20K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 12K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Facebook
Twitter