LinkBack URL
About LinkBacksRead more.Optimisation reference manual provides tech insights into recent Intel Pentium Silver CPUs.
Read more.Optimisation reference manual provides tech insights into recent Intel Pentium Silver CPUs.
specs seem like it might be impressive.
You don't get a lot of detail on these block diagrams, but it doesn't look/sound any better than an old A57 ARM core. The 3 decode/4 issue sounds quite poor compared to an A72 which is 3 decode/8 issue.Originally Posted by ETR316
I imagine a custom core like Apple's will wee all over this, yet Intel will try and charge a premium.
Hard to gauge performance in that manner because internal micro operations for each architecture are quite different. ARM's Cortex-A72 is only near Intel's Apollo Lake performance level, for example Google's OP1 (Rockchip RK3399) versus Intel's Pentium N4200 https://browser.geekbench.com/v4/cpu...seline=5344844 although Intel's Pentium N4200 is much faster under Android operating system https://browser.geekbench.com/v4/cpu...seline=4531039 However Intel's newer Gemini Lake is much more faster, for example Google's OP1 (Rockchip RK3399) versus Intel's Pentium N5000 https://browser.geekbench.com/v4/cpu...seline=5315637 Note that Intel Pentium N5000 was under Microsoft Windows operating system (but should be much faster under Linux and Android operating systems).
Apple's own custom ARM cores are only used for their own iPhone and iPad products. Thus not exactly in the same market sector. These Intel Gemini Lake SoCs are meant for low cost devices like cheap laptops, 2-in-1 hybrids, mini PCs and Chromebooks. Also the listed prices does not reflect the actual pricing ODMs get.
Last edited by LordRetroGamer; 30-12-2017 at 02:51 PM.
And yet the now rather old and A57 based Tegra X1 does better despite having what is clock for clock a worse core: https://browser.geekbench.com/v4/cpu...seline=5344844
which shows us once again that although system level benchmarks are all we have, they aren't useful for discussing a component performanceThe 1.1GHz rating of the N4200 there is probably a miss-representation. A laptop that size with a 6W APU, I would expect it to be on permanent 2.5GHz boost.
As for the Apple chip not being in the same market, I beg to differ. Intel would adore having their CPUs in Apple handhelds, or frankly in anyone's handhelds. However, even as someone who doesn't like Apple products I have to give them the nod for a core well done, so I don't see Intel ever getting that gig.
Should be noted that NVIDIA Shield Android TV had a cooling fan http://www.neoseeker.com/Articles/Ha...roid-tv/3.html thus less thermal throttling than fanless laptops (like low cost Chromebooks). The old ARM Cortex-A57 inside NVidia's Tegra X1 is only around Intel's Silvermont performance https://browser.geekbench.com/v4/cpu...seline=2086522 However Intel's Goldmont core in Intel's Apollo Lake chips are a step faster than both ARM Cortex-A57 and Intel's Silvermont cores. Most x86 based architecture can run at higher frequencies then ARM based architectures. For reference https://ark.intel.com/products/95592...-up-to-2_5-GHz that 2.5GHz is single core boost frequency while 1.1GHz is base frequency with all cores utilized since these SoCs are typically used in fanless (passively cooled) laptops. Thus 6W is the upper worse case limit in such devices where thermal throttling is always expected.And yet the now rather old and A57 based Tegra X1 does better despite having what is clock for clock a worse core: https://browser.geekbench.com/v4/cpu...seline=5344844
which shows us once again that although system level benchmarks are all we have, they aren't useful for discussing a component performance
Furthermore Apple charges premium prices for those products (iPhone and iPads), which differs from the low budget market sectors where Intel's Apollo Lake (and Gemini Lake) SoCs occupy. As for handhelds with Intel's SoCs, there are lots of examples like Vastking G800 (with Apollo Lake https://www.youtube.com/watch?v=6Da3otMXWxI ), upcoming GPD Win 2 (with Kaby Lake https://www.youtube.com/watch?v=21AsXQwfxoc ), current GPD Win (with Chery Trail https://www.youtube.com/watch?v=7C51mkucrnc ), GPD Pocket (with Cherry Trail https://www.youtube.com/watch?v=BkQPTXB3DaE ), Linx Vision 8 Gaming Tablet (with Chery Trail https://www.youtube.com/watch?v=r160OjjTUAk ), Gole1 (with Cherry Trail https://www.youtube.com/watch?v=oJTFnqZqnZQ ), Leagoo T5c (with Spreatrum SoC featuring Airmont cores http://www.leagoo.com/product/t5c/ ), etc.
Last edited by LordRetroGamer; 31-12-2017 at 02:37 AM.
MLyons (02-01-2018)
Go back and look at the scores you linked to, the multi core score is 3X higher than the single core score which doesn't make sense if single core is 2.5GHz and multi core is 1.1GHz, that would work out at 1.1x4/2.5=1.76 times faster in multi core if you could get perfect scaling which ofc you never do. I can't see 6W in a 15.6" laptop chassis needing a fan, my wife's Tegra based Chromebook certainly doesn't.
I am aware that Atom devices exist, I have quite a few around the house. IME they seem like a really good idea until you try to actually use it for something. My son wanted one to play Minecraft on, but a Z3735D in a 10" 2-in-1 isn't even up to that job. Perhaps my daughter's Cherry Trail device would do a better job (I gather it has dual channel ram for starters) but she gave up on it pretty fast so who knows.
Bay Trail die size was just over 100mm2, same sort of size that Apple seem to aim for with their SoC die. Except Intel's advanced process boosts not just the performance but also the cost per square mm. So they *should* have been in comparable devices.
Edit: I think my biggest beef here is that these Intel releases just bore me. It is the last mediocre part made slightly better, but not too good because that might step on the lucrative desktop parts so best cripple it a bit eh?
Last edited by DanceswithUnix; 31-12-2017 at 10:44 AM.
Welcome to Intel Corp 'progress'. This is why I would be happy to never buy another Intel product. And they STILL haven't paid EU fines afaik.
hexus trust : n(baby):n(lover):n(sky)|>P(Name)>>nopes
Be Careful on the Internet! I ran and tackled a drive by mining attack today. It's not designed to do anything than provide fake texts (say!)
Frequency across all cores depends very much on heat dissipation of the device. There is always some thermal throttling in passively cooled devices. Heck, even with the highest single core result https://browser.geekbench.com/v4/cpu/5439077 doesn't mean multicore result will be the highest as well (also shows that previous single core result was not always running at highest boost frequency). Throttling does not mean always running at base frequency, rather when reaching worse case thermal limits then will drop down to base frequency however when thermal headroom becomes available then will increase operating frequency. As mentioned earlier, thermal throttling is always expected. Also NVIDIA Tegra based Chromebooks used the much lower power NVIDIA's Tegra K1 (useable with passive cooling) instead of the higher power NVIDIA's Tegra X1 (which active cooling is typically recommended).
There are always some configuration, modifications or patches (like Optifine ) to allow Minecraft to run decently on Intel Atom-based devices. Example https://www.youtube.com/watch?v=7C51mkucrnc&t=6m38s That is well known already.
Originally Intel's Silvermont cores was squarely aimed to compete with ARM's generic Cortex-A57 cores, thus was not competing with Apple's custom SoCs in Apple's own devices. Intel's Avoton for microservers was also using Intel's Silvermont cores and competes with ARM-based server SoCs with ARM Cortex-A57 cores such as AMD's Seattle (incidentally the often delayed AMD's Seattle seems to have dropped off the microserver scene not long after it was finally launched).
These SoCs are meant for specific markets, mainly low cost sectors (e.g. Chromebooks) and embedded (e.g. IoT). Thus only good reason to update its performance is to keep up with ARM's latest generic cores (e.g. ARM's Cortex-A73).
Last edited by LordRetroGamer; 01-01-2018 at 02:33 AM.
MLyons (02-01-2018)
Yes, but the great irony (and source of great Schadenfreude) is that precisely because Intel spent so many years defending their cash cow iCore products and neglecting* Atom and mobile below laptops, that they missed the rise of mobile and are have spend the last few years frantically attempting to re-enter mobile.
However, even throwing $billions into it by dumping (well Intel PR would call it contra-revenue) hasn't made any difference. Firstly because ARM is currently still 'better', and secondly because Intel is very reluctant to make custom or semi-custom designs (EMIB might eventually alter that) especially if they could compete with more profitable iCore part and thirdly (and probably the main reason) is that the major mobile vendors know precisely how they and the other OEMs would be treated if Intel ever gain a monopoly on mobile. After all, a lot of the mobile OEMs also produce computer parts where they've mostly learned to fear Intel (both for their tactics and the fact that Intel take almost all of the margins of the finished parts).
*Neglecting is of course a charitable term. Crippling it so that Atom doesn't affect sales of more expensive iCore parts is closer to that they actually did.
Haha kompukare that warms my soul as 2017 draws to an end. How can any company, large or small, operate so far outside of common moral value / state ideology and still expect things to turn out rosy in the end? What was their goal? Was it to actually create a valuable value add entity or just to entitle a bunch of idiots with nice job titles and shady share options or whatever.
man the capitalistic economy baffles me sometimes. There's 10x more value in your average corner grocer than there is in some of these nihilistic startup wannabes. May they go the hell away.
hexus trust : n(baby):n(lover):n(sky)|>P(Name)>>nopes
Be Careful on the Internet! I ran and tackled a drive by mining attack today. It's not designed to do anything than provide fake texts (say!)
Yes, Intel missed the early rise of tablets and smartphones because they didn't have a suitable contender yet (as can be seen from early Moorestown attempt). Only much later did Intel have contenders starting with Cloverview/Medfield and Clover Trail, then followed by Moorefield and Bay Trail. However Intel left the market due to competition (from ARM based SoCs), pricing pressure (includes margins) and losses from contra revenues (used in promoting Intel's Atom SoCs). Intel's business model could not cope with all of that. Nowadays the business model in this sector have changed, shifting towards semi-custom (e.g. Spreadtrum's 8-Core Airmont SoC).
There is still a difference however (even after contra revenues have ended). Because that contra revenue push came along with full technical and engineering support from Intel, now nearly every Chinese OEM/ODM became acclimatized with Intel's chips and can produce any type of device almost at will. This results in a sudden surge of numerous "no name" brands (e.g. Chuwi, Teclast, Onda, ALLDOCUBE, Jumper, Pipo, etc) taking on the low budget market for notebooks, 2-in-1 hybrids and tablets. This also led to bigger names (in China) like Xiaomi and Huawei entering the notebook market. Additionally its not surprising that relatively unknown GPD could suddenly make UMPCs (like GPD Win, GPD Pocket and soon GPD Win 2). Heck, I have not heard of Vastking until that G800 gaming tablet appearance at a tradeshow first reported by Notebook Italia.
Firstly its difficult to compete against cheaper ARM based SoCs (from Mediatek, Rockchip, Allwinner, AmLogic, etc). Google's Pixel C tablet recent demise joins a long casualty list. Secondly there were some Intel semi-custom designs that you are not aware of. Intel's Avoton (Atom C2000 series) and Broadwell-DE (Xeon D series) were actually the result of Facebook's collaboration. There's also Intel and Rockchip partneship's doomed Sofia (using TSMC's 28nm). Spreadtrum's SC9861G and SC9853I SoCs were using Intel's Airmont cores and built on Intel's 14nm process technology. Thirdly those OEMs/ODMs continued their business as usual of still producing ARM powered devices (e.g. tablets, smartphones, etc) along with extra business of producing Intel powered devices. Its all about market demands and business sense. Why do you think the market is still full of cheap ARM powered tablets and smartphones? And nothing to do with your claims on "fear", "tactics" or "margins". Intel could not make much margins in this area of the market, thus inevitably ending further development of their ultramobile Atom line.
Different markets and different prices. Intel's more expensive Core i and M series have larger CPU core area (for that extra performance). Also do you realize that Intel's current (Kaby Lake) Core M series main chip (with dual core CPU and integrated GPU) die size is already around 100mm^2, which is already larger than the die size of Intel's (quad core) Atom SoC chip? That does not include the external PCH chip (in MCM package with total area of 338mm^2). In contrast, Intel's (quad core) Atom SoC has integrated PCH and more I/Os in the same chip. For your information, Intel's Cherry Trail die size is estimated to be around 83mm^2 only. The smaller die size of Intel's Atom series meant it could be sold cheaper. Many cheap ARM based SoCs have die sizes way under 100mm^2 and still made with older 28nm planar process (which had much lower cost than newer FinFET process).
Last edited by LordRetroGamer; 01-01-2018 at 02:23 AM.
MLyons (02-01-2018)
LordRetroGamer I just wanted to chime in to say thanks for the level of detail of your posts and the explanatory nature. You may like PCPER podcast, at least one of the presenters is well versed in process tech and CPU architecture. And that seems right up your street
hexus trust : n(baby):n(lover):n(sky)|>P(Name)>>nopes
Be Careful on the Internet! I ran and tackled a drive by mining attack today. It's not designed to do anything than provide fake texts (say!)
View the profiles of people named CoCo Spy. Click here and Join Facebook to connect with CoCo Spy and others you may know.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Reply With Quote