Page 43 of 78 FirstFirst ...
33
41
42
43
44
45
53
... LastLast
  1. #841
    Quote Originally Posted by Evildeffy View Post
    Any electrical engineer would turn around in his grave reading this statement.
    No, he wont. Both use multicore structure with integrated FPUs, both are using three level cache structure. There are differences but that doesnt make them completely different designs.

    Quote Originally Posted by Evildeffy View Post
    No, now you're altering things, you specifically stated that Ryzen and Bulldozer are very close to each other and that software control and better lithography attributed to it.
    The uArch itself, and it is VASTLY different from Bulldozer to Ryzen, is the reason for power efficiency as well and stating that not to be the case is ... just WTF.
    Quoting myself:
    I dont see much improvement in power consumption in Zen over Buldozer due to architecture changes
    I guess you cannot read. Yes, I think that litography and software control are the most significant factors in power efficiency improvement in Zen over Bulldozer. Zen looks to be exactly what AMD had in mind while designing Bulldozer but couldnt get the most of it with the level of technology they had at the time.

    Quote Originally Posted by Evildeffy View Post
    You have not had a sweeping change from Skylake -> Kaby Lake and will not with Coffee Lake as it is yet again a refinement of Kaby Lake.
    I think so too but for now we have zero concrete data. IPC improvement also seem very significant (largest so far since Sandy Bridge, and that's considering both Kaby and Coffee Lake both use 14nm nodes).

    Quote Originally Posted by Evildeffy View Post
    That they are both based upon the x86 instruction set doesn't mean they can't go about it different ways.
    Sure, but it's not like AMD hasnt been going that "modularity" and core clustering way since Bulldozer. Again, comparing the two makes you wonder how much more similar those two could've been if AMD had better tech back then.

    Quote Originally Posted by Evildeffy View Post
    The Internal Combustion Engine has the same principles and yet can drive on a multitude of different fuels provided they have a different design.
    Diesel vs. Benzine (Petrol if you're British/US) vs. Water vs. Natural Oils.
    Same principle ... entirely different execution.
    I agree, but I'd say that it's not comparing gasoline to diesel engines, it's like comparing 80s diesel with a modern one.

    Quote Originally Posted by Evildeffy View Post
    And how did they manage to do that? That's right they managed to separate the iGPU control entirely from the CPU.
    That, which is impossible now, is why they could and Intel having fixed the "bug" by locking iGPU control with the rest of the CPU again means you'll never see this again, meaning ... in EVERY definition ... iGPU is most definitely present in CPU Power Efficiency and there's no way to dodge this.
    Also Kaby Lake's "Intel Management Engine" is not OS based, it is most assuredly based in the firmware of the board, this isn't part of the OS otherwise it'd be 100% circumvented already.
    iGPUs are powered by a different VRM on the board and you can actually check that: just disable the iGPU and stab those mosfets with a multimeter. The only reason why Intel doesnt allow BCLK overclocking is to avoid cannibalizing their own K processors. Also, yes, iGPUs are affected by BCLK overclocking but there is a standard feature on all Intel 100 and 200 series boards: iGPU is automatically disabled when you raise BCLK, it's also automatically disabled if you have discrete GPU.

    Quote Originally Posted by Evildeffy View Post
    Across different uArch and lithography I would agree with you.
    Staying on the same uArch and Lithography it's process ... it's actually VERY viable to extrapolate some things such as power draw/envelope.
    So, Ryzen 7 1700 and Ryzen 7 1700X? What do you extrapolate here? Which one is cheating?

    Quote Originally Posted by Evildeffy View Post
    No 2 CPUs are identical, you have higher quality silicon (natural part of the process), speed and operating voltages at said speed.
    1 may be extremely efficient requiring lower voltages than the other, center silicon wafer, and the other may come from the edges.
    1 may be clocked considerably higher and require more operating voltage than the other.
    You may have 1 which is inefficient and requires more operating voltage as well as being clocked higher with speeds, which is what AMD bases it's TDP upon the maximum voltage x electrical current required for operation and cooling to associate with that.
    There's a plethora of reasons of why and how but let's get to that shall we as with your very example.
    Weren't you the one who stated the following?

    So what is it now then as you first say it is based upon heat dissipation and now not.

    Even though this is a power virus so consumption goes beyond stated TDP as ordinary software cannot generate the same thing:
    Click me for a picture!
    Or are you saying these readings do not properly convey the differences since you don't believe there's a difference?
    Also check out the 7700K there ... higher power draw than a 95W TDP CPU whilst itself is 91W ... strange ... the difference mustn't be in the uArch or anything, it must be because it's got no solder under the hood.

    Do you think there would be a meaningful change to power draw and cooling requirements if you were to repeat the same test with a soldered/delidded CPU?

    Also still dodging... yes you are very much lacking.
    It's you who is dodging here. Just answer the question. Why does the same silicon have a 30W lower TDP and it's apparently ok, but Intel cannot possibly keep their new processor under 100W TDP. We are not discussing how right/honest TDP ratings are, because they are not.

    And yes, delidding/soldering would actually decrease cooling requirements (wont do anything for the power draw though).
    R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B

  2. #842
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Thunderball View Post
    No, he wont. Both use multicore structure with integrated FPUs, both are using three level cache structure. There are differences but that doesnt make them completely different designs.
    And yet again this statement shows an EE would turn in his grave.
    Going back to the engine analogy: Diesel -> Petrol.
    Just because it has cylinders doesn't mean the engine is identical.

    Quote Originally Posted by Thunderball View Post
    Quoting myself:

    I guess you cannot read. Yes, I think that litography and software control are the most significant factors in power efficiency improvement in Zen over Bulldozer. Zen looks to be exactly what AMD had in mind while designing Bulldozer but couldnt get the most of it with the level of technology they had at the time.
    Ryzen is a complete departure from Bulldozer as a uArch... yes AMD has learned things (painfully) with Bulldozer but they aren't even remotely alike.
    That's literally like saying a Peugeot 206's 1.4 benzine engine is the same as the Bugatti Veyron ... it's not even close.

    Quote Originally Posted by Thunderball View Post
    I think so too but for now we have zero concrete data. IPC improvement also seem very significant (largest so far since Sandy Bridge, and that's considering both Kaby and Coffee Lake both use 14nm nodes).
    It's a refinement of the same uArch as Skylake, it's an "Optimization" as Intel calls it so there will not be any sweeping changes as a fact and so far the leaks (which I posted here) state 11% difference in Single Core performance.
    There's a 4,4% gain in clock speed for Single Core Turbo which means ~5% IPC improvement ... it's not the biggest since Sandy Bridge at all.
    It's just the same generational baby step ... but this is counted by "Intel's SPEC" something .. their own benchmark that caught them cheating in the past just for owning a CPU that wasn't theirs (CPU scored lower in the exact same benches just for not being IDed as GenuineIntel).
    So that very well remains to be seen.

    Quote Originally Posted by Thunderball View Post
    Sure, but it's not like AMD hasnt been going that "modularity" and core clustering way since Bulldozer. Again, comparing the two makes you wonder how much more similar those two could've been if AMD had better tech back then.
    They aren't even remotely similar to each other, I don't even know how you can say that.
    The "CCX" are a means for scaling CPUs, they aren't even in the same universe as the "modules" ... Please tell me how the uArchs are even remotely similar.

    Quote Originally Posted by Thunderball View Post
    I agree, but I'd say that it's not comparing gasoline to diesel engines, it's like comparing 80s diesel with a modern one.
    You're seriously frustrating, I honestly don't know what the hell goes through your head.

    Quote Originally Posted by Thunderball View Post
    iGPUs are powered by a different VRM on the board and you can actually check that: just disable the iGPU and stab those mosfets with a multimeter. The only reason why Intel doesnt allow BCLK overclocking is to avoid cannibalizing their own K processors. Also, yes, iGPUs are affected by BCLK overclocking but there is a standard feature on all Intel 100 and 200 series boards: iGPU is automatically disabled when you raise BCLK, it's also automatically disabled if you have discrete GPU.
    If only that were truly the case, it'd mean every CPU would be overclockable.
    The iGPU unfortunately doesn't turn off when you BCLK OC, it is the very reason why you can only get a few MHz out of BCLK as long as that's enabled.
    Whether there's full load on the GPU or not is another matter, where do you think the load is when you use QuickSync?
    It simply "disables" the registry microcode from showing it as a device in... it's still active.
    I suggest you read exactly how SuperMicro managed to get that trick to using BCLK OC to work on consumer Intel CPUs and why HEDT CPUs still can be OCed by BCLK without issues.

    Quote Originally Posted by Thunderball View Post
    So, Ryzen 7 1700 and Ryzen 7 1700X? What do you extrapolate here? Which one is cheating?
    Nice, cutting off the answer to suit your needs, luckily you still quoted it in your next quote for your answer.
    No cheating, pure logic.

    Quote Originally Posted by Thunderball View Post
    It's you who is dodging here. Just answer the question. Why does the same silicon have a 30W lower TDP and it's apparently ok, but Intel cannot possibly keep their new processor under 100W TDP. We are not discussing how right/honest TDP ratings are, because they are not.

    And yes, delidding/soldering would actually decrease cooling requirements (wont do anything for the power draw though).
    I have answered it multiple times but you either ignore it or are unable to understand it.
    Read through the posts again, who knows .. maybe you'll actually read and understand it at some point.

    Keep dodging and selective reading/quoting though, who knows you may get lucky and I won't notice you doing it.

    Also ... decreasing cooling requirements doesn't decrease TDP if the power consumption figures don't change meaning neither does cooling requirements.
    Bust your head on that for a bit.
    Last edited by Evildeffy; 2017-08-19 at 12:23 AM.

  3. #843
    Warchief
    10+ Year Old Account
    Join Date
    Feb 2011
    Location
    Scottishlands
    Posts
    2,035
    Fuck Intel.

    The 270 chipset is a year old and it's being abandoned for no reason. Coffee Lank CPUs could easily work with 270 platform. There is no need to have a new platform every year, it's just pure greed. Once my 4790K dies, I'm moving over to AMD. At least then I'll know my mobo will be supported for a good few years if I wish to upgrade the CPU. Right now if you want to upgrade on Intel, you have to buy a new mobo and cpu. I'd rather spend good money on a mobo that'll last me at least 4 to 6 years, and all I need to do is drop in a new CPU when I want to upgrade. I'll take the single core performance loss tbh.

  4. #844
    Quote Originally Posted by Tommi View Post
    Fuck Intel.

    The 270 chipset is a year old and it's being abandoned for no reason. Coffee Lank CPUs could easily work with 270 platform. There is no need to have a new platform every year, it's just pure greed. Once my 4790K dies, I'm moving over to AMD. At least then I'll know my mobo will be supported for a good few years if I wish to upgrade the CPU. Right now if you want to upgrade on Intel, you have to buy a new mobo and cpu. I'd rather spend good money on a mobo that'll last me at least 4 to 6 years, and all I need to do is drop in a new CPU when I want to upgrade. I'll take the single core performance loss tbh.
    Who says that it could easily work with the 270 platform? All rumors i hear is that due to the fact that 2 cores are added they need to use a different pinout and therefore current 270 platform boards cannot be compatible due to not having the correct pinout for Coffee Lake as they are designed for Kabylake.

  5. #845
    Quote Originally Posted by Evildeffy View Post
    If only that were truly the case, it'd mean every CPU would be overclockable.
    The iGPU unfortunately doesn't turn off when you BCLK OC, it is the very reason why you can only get a few MHz out of BCLK as long as that's enabled.
    Whether there's full load on the GPU or not is another matter, where do you think the load is when you use QuickSync?
    It simply "disables" the registry microcode from showing it as a device in... it's still active.
    I suggest you read exactly how SuperMicro managed to get that trick to using BCLK OC to work on consumer Intel CPUs and why HEDT CPUs still can be OCed by BCLK without issues.
    Stop spreading bullshit. Just pick a multimeter and stab your iGPU phase, it cannot be active if it doesnt have power. I, for one, dont have an option to use QuickSync, because my iGPU is disabled. I know how SuperMicro did that, but Intel patched that trick on Kaby Lake: OS simply wont boot. You can still overclock K chips through BLCK if your motherboard has specific hardware, same with Ryzen btw (ASRock, for one, has that on their top Z270 boards).


    Quote Originally Posted by Evildeffy View Post
    Nice, cutting off the answer to suit your needs, luckily you still quoted it in your next quote for your answer.
    No cheating, pure logic.
    Havent seen any answers on Ryzen, quote it. And yes, it's a completely valid comparison, same hardware, 30W TDP difference. Yes, clocks are different but those dont amount to nowhere close to 30W, in any case.

    Quote Originally Posted by Evildeffy View Post
    I have answered it multiple times but you either ignore it or are unable to understand it.
    Read through the posts again, who knows .. maybe you'll actually read and understand it at some point.
    You havent, all I keep hearing is "different uarch, different litography". Ryzen 7 1700 and Ryzen 7 1700X are the same chip configured slightly differently.

    Quote Originally Posted by Evildeffy View Post
    Also ... decreasing cooling requirements doesn't decrease TDP if the power consumption figures don't change meaning neither does cooling requirements.
    Bust your head on that for a bit.
    Heat dissipation =/= power consumption, and as TDP is a rating meant for cooling suppliers, so it would most likely change.
    R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B

  6. #846
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Quote Originally Posted by Thunderball View Post
    No, he wont. Both use multicore structure with integrated FPUs, both are using three level cache structure. There are differences but that doesnt make them completely different designs.
    Holy shit, just stop. Please, just stop.
    Read starting here for 4 pages
    http://www.anandtech.com/show/11170/...00x-and-1700/6
    and this
    http://www.anandtech.com/show/10578/...archy-revealed

    and look at these


  7. #847
    I dont get it though, why is this whole TDP discussion so important. Seeing talk seems mostly about K cpu's, when overclocking TDP goes out of window anyway. And if your not overclocking with a K cpu, your basicly throwing away money.

    Then again that being said, i see no reason not to believe that Coffee lake's power consumption at stock will be under TDP for most workloads (Not looking at stuff like Prime 95 FFT's or stuff with AVX in the mix that will break TDP on (nearly) every cpu). In most if not all of tests the 7700K and 7700 when running a real work workload stay well below their rated TDP (out of my head for example at Tom's the 7700K consumes like 46W during a autocad Workload and 77W during a gaming loop. That should leave quite some room for Intel to tune the 6 core boost speed in such a way that it stays under TDP when used at stock settings and no shenanigans from mobo manfacturors are active in the UEFI that mess with the stock boost bin's.
    Last edited by chronia; 2017-08-19 at 08:21 AM.

  8. #848
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Thunderball View Post
    Stop spreading bullshit. Just pick a multimeter and stab your iGPU phase, it cannot be active if it doesnt have power. I, for one, dont have an option to use QuickSync, because my iGPU is disabled. I know how SuperMicro did that, but Intel patched that trick on Kaby Lake: OS simply wont boot. You can still overclock K chips through BLCK if your motherboard has specific hardware, same with Ryzen btw (ASRock, for one, has that on their top Z270 boards).
    Why don't you go ahead and try it then, see if it works.
    The entire trick to use it, which is why it doesn't work anymore, was locked up because of them not wanting overclocking on non-K models.
    And it's done by involving the iGPU because that's the simplest way to go about it as the iGPU simply doesn't like it being pushed that way at all.

    But you know what, you seem to think it's still possible... go ahead .. grab a non-HEDT Intel CPU and drop it in and try proper base clock overclocking, let's see how far you get.
    Also like I've mentioned already (which you skip to try and strengthen your point) is that HEDT chips and Ryzen (if you have an external clock generator on the mobo) can use Base Clock overclocking because they have NO iGPU to stop it from going haywire.

    Quote Originally Posted by Thunderball View Post
    Havent seen any answers on Ryzen, quote it. And yes, it's a completely valid comparison, same hardware, 30W TDP difference. Yes, clocks are different but those dont amount to nowhere close to 30W, in any case.
    I most certainly have, not my problem you're selectively reading or simply cannot understand why, try again, use a microscope to read every letter, it's in there MULTIPLE times as a direct answer.

    Quote Originally Posted by Thunderball View Post
    You havent, all I keep hearing is "different uarch, different litography". Ryzen 7 1700 and Ryzen 7 1700X are the same chip configured slightly differently.
    Identical answer as above seeing as you're folding the same part into 2 questions to again further your own goals which continue to demonstrate your lacking knowledge.

    Quote Originally Posted by Thunderball View Post
    Heat dissipation =/= power consumption, and as TDP is a rating meant for cooling suppliers, so it would most likely change.
    I'll allow you to think a little bit on what you've said here and then re-think the reason for TDP existence, who knows you might actually learn something.

    Now you also have others telling you the difference between 2 differing uArchs and them being basically incomparable, remember that.

    - - - Updated - - -

    Quote Originally Posted by chronia View Post
    I dont get it though, why is this whole TDP discussion so important. Seeing talk seems mostly about K cpu's, when overclocking TDP goes out of window anyway. And if your not overclocking with a K cpu, your basicly throwing away money.

    Then again that being said, i see no reason not to believe that Coffee lake's power consumption at stock will be under TDP for most workloads (Not looking at stuff like Prime 95 FFT's or stuff with AVX in the mix that will break TDP on (nearly) every cpu). In most if not all of tests the 7700K and 7700 when running a real work workload stay well below their rated TDP (out of my head for example at Tom's the 7700K consumes like 46W during a autocad Workload and 77W during a gaming loop. That should leave quite some room for Intel to tune the 6 core boost speed in such a way that it stays under TDP when used at stock settings and no shenanigans from mobo manfacturors are active in the UEFI that mess with the stock boost bin's.
    It's not about the TDP in general but the consequences that come with it when you go from Architecture A to B whilst remaining on the same lithography process and uArch design.
    Of course most of the time you cannot use Autodesk AutoCAD (notoriously poorly optimised, 1 thread only) nor gaming in general and reach your TDP.
    In fact a great deal of CPUs will be just under their rated TDP when used under full load that isn't a power virus (like Prime95), but when utilized fully it will draw close to it's rating on tasks that stress the CPU on a normal job.

    However all that doesn't change the "minimum behaviour" Intel sets forth if the leaks are true with a 4,3GHz all-core boost.
    Because for all intents and purposes that 4,3GHz becomes "Base Clock" as the only way to not use this is if you're overheating due to insufficient cooling, in which case you already have other problems, or underclock it.
    And to overcome and remain within it's TDP means that the old Kaby Lake uArch needs to improve by a considerable amount in order to accommodate these changes since Coffee Lake is a refinement of Kaby Lake, not a new process or Architecture, and therein lies the issue... the amount that is required to be improved is of such a degree that it's unlikely to happen without a completely different uArch or Lithography Process.

    Hence why I'm stating that Intel is outright "lying/misleading" with it's power envelope numbers for this 6-core.
    Some people may not care at all and for understandable reasons... I wouldn't give 2 flying fucks either about it either if I wanted it.

    But it does hold up to the point that Intel is up to shenanigans in order to paint their position better than it really is.
    We've already seen Intel shit-sling at AMD for their ThreadRipper platform of "4 desktop dies glued-together" and "Lack of ecosystem" etc...
    It's just the consumer friendly-ness at stake along with the fact that someone may purchase a said 8700K f.ex. and buy a cooler that just barely fits the 95W TDP bill and gets throttling as a result because of Intel's "TDP specs!".

  9. #849
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Xinkir View Post
    So is the 8700k supposed to be better single thread performance too and overclock-ability then the 7700k or nah? With more cores I'd assume base clock is less and probably can't hit 5.0 or higher like the 7700k can. I mean I'm only asking on 2-3 cores it doesn't have to be all cores. 21st is also the eclipse which seems like a weird day to have an announcement.
    According to leaks the 8700K is supposedly 11% faster, but this is measured by Intel's own software whom claimed the 7700K was 15% faster than the 6700K as well.
    So we'll have to see on reviews/release whether that's true or not.

    As for overclockability... possibly and possibly not, the silicon got considerably more complex and larger and the general rule of thumb is that if you make something more complex it should clock lower... so until we get reviews that aren't golden sample based (like the Skylake-X launch) we simply do not know.

    Currently "leaked" turbo for 2 cores is 4,6GHz which is 100MHz faster than the 7700K All-Core turbo, whether this remains true is also something we'll know during either launch or announcement, depending on if clocks are final.

    As far as eclipse during announcement.... ILLUMINATI CONFIRMED?!?!

  10. #850
    Dreadlord Enfilade's Avatar
    10+ Year Old Account
    Join Date
    May 2010
    Location
    Pennsylvania
    Posts
    953
    8700K will be stable at 5.0 GHz on all cores. You heard it here first.

  11. #851
    Quote Originally Posted by Remilia View Post

    That's not core architecture though, it's a pipeline diagram.
    R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B

  12. #852
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Quote Originally Posted by Thunderball View Post
    That's not core architecture though, it's a pipeline diagram.
    It's called a CPU block diagram. That is how a singular core works for those architectures.
    Or just read this entire page https://en.wikichip.org/wiki/amd/microarchitectures/zen
    Last edited by Remilia; 2017-08-20 at 01:02 AM.

  13. #853
    Quote Originally Posted by Evildeffy View Post
    Why don't you go ahead and try it then, see if it works.
    The entire trick to use it, which is why it doesn't work anymore, was locked up because of them not wanting overclocking on non-K models.
    And it's done by involving the iGPU because that's the simplest way to go about it as the iGPU simply doesn't like it being pushed that way at all.

    But you know what, you seem to think it's still possible... go ahead .. grab a non-HEDT Intel CPU and drop it in and try proper base clock overclocking, let's see how far you get.
    Also like I've mentioned already (which you skip to try and strengthen your point) is that HEDT chips and Ryzen (if you have an external clock generator on the mobo) can use Base Clock overclocking because they have NO iGPU to stop it from going haywire.
    It's impossible to do with Kaby Lake - Intel patched the Management Engine: OS wont boot with a BCLK overclock. You can still overclock it (if you have an external clock generator) and watch it go in UEFI though =) It's still physically possible, just locked through low level drivers. I've had some fun with an i3-6100 though.

    This discussion is getting out of topic though: you can absolutely disable the iGPU because it actually has a separate VRM. Also, for example on Sandy Bridge and Haswell you could actually overclock an iGPU, even on non-Z boards (but only on K SKUs in case of Haswell). Introducing separate base clock generators for everything (mem. controller, igpu, CPU cores) makes chipsets and the boards they're using that much expensive though. By the way, it's interesting too see how AMD will circumvent memory controller problems on Ryzen APUs: right now memory controllers are using a separate VRM but that same VRM is also designed to handle APUs, and AMD APUs are not lowsy.

    Quote Originally Posted by Evildeffy View Post
    However all that doesn't change the "minimum behaviour" Intel sets forth if the leaks are true with a 4,3GHz all-core boost.
    Because for all intents and purposes that 4,3GHz becomes "Base Clock" as the only way to not use this is if you're overheating due to insufficient cooling, in which case you already have other problems, or underclock it.
    Ryzen doesnt work that way, for example. You're not going to see all cores boosting at the same time, even under full load. You can actually see that pretty clearly on Threadripper: overclocking past the base boost clock (but below XFR clock, obviously) actually reduces performance in most multithreaded workloads compared to stock. That's pure software control. Intel probably didnt have any incentive to do something like that in the past, but here is one.

    Quote Originally Posted by Evildeffy View Post
    And to overcome and remain within it's TDP means that the old Kaby Lake uArch needs to improve by a considerable amount in order to accommodate these changes since Coffee Lake is a refinement of Kaby Lake, not a new process or Architecture, and therein lies the issue... the amount that is required to be improved is of such a degree that it's unlikely to happen without a completely different uArch or Lithography Process.
    Again, that doesnt need to be architecture or litography.

    Quote Originally Posted by Evildeffy View Post
    But it does hold up to the point that Intel is up to shenanigans in order to paint their position better than it really is.
    We've already seen Intel shit-sling at AMD for their ThreadRipper platform of "4 desktop dies glued-together" and "Lack of ecosystem" etc...
    It's just the consumer friendly-ness at stake along with the fact that someone may purchase a said 8700K f.ex. and buy a cooler that just barely fits the 95W TDP bill and gets throttling as a result because of Intel's "TDP specs!".
    Well those Threadripper comments sounded totally unprofessional, but from a logical standpoint they are absolutely true.

    You can try slapping a 95W rated cooler on R7 1800X or 7700K right now, see how it goes. Gonna be the same result.
    R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B

  14. #854
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Thunderball View Post
    It's impossible to do with Kaby Lake - Intel patched the Management Engine: OS wont boot with a BCLK overclock. You can still overclock it (if you have an external clock generator) and watch it go in UEFI though =) It's still physically possible, just locked through low level drivers. I've had some fun with an i3-6100 though.

    This discussion is getting out of topic though: you can absolutely disable the iGPU because it actually has a separate VRM. Also, for example on Sandy Bridge and Haswell you could actually overclock an iGPU, even on non-Z boards (but only on K SKUs in case of Haswell). Introducing separate base clock generators for everything (mem. controller, igpu, CPU cores) makes chipsets and the boards they're using that much expensive though. By the way, it's interesting too see how AMD will circumvent memory controller problems on Ryzen APUs: right now memory controllers are using a separate VRM but that same VRM is also designed to handle APUs, and AMD APUs are not lowsy.
    On official BIOSes it's plain impossible.
    iGPU still prevents BCLK overclocking and the silicon of the GPU is still powered even if the VRM are humming along at almost 0 load.
    You can still overclock an AMD APU (Kaveri/Godavari etc.) using base clocks if there's an external clock generator, still not possible on Intel's official BIOS and MicroCode.
    You can "disable" the iGPU all you want, it's still powered and tied into the whole system bus, SuperMicro found a vulnerability to detach it from the bus clock and BCLK overclocking was reborn ... for a short while.

    Go back to every HEDT platform and the CPUs prior (including even the Xeons, Core2 and Pentiums) which could ALL overclock with base clock changes, this is still possible because there's no iGPU tied into it, but w/e ... you clearly seem to change the subject to avoid talking about how wrong you were.

    AMD's Vega is actually incredibly efficient on lower clock speeds so their APUs will likely be the same.

    Quote Originally Posted by Thunderball View Post
    Ryzen doesnt work that way, for example. You're not going to see all cores boosting at the same time, even under full load. You can actually see that pretty clearly on Threadripper: overclocking past the base boost clock (but below XFR clock, obviously) actually reduces performance in most multithreaded workloads compared to stock. That's pure software control. Intel probably didnt have any incentive to do something like that in the past, but here is one.
    Ryzen 7's All-Core boost is 3,7GHz and multi-threaded tasks still boost when all cores are manually OCed to 4GHz compared to stock, it's Single Core Boost (XFR) which is faster than such an OC.
    Intel's silicon/uArch/OC doesn't work that way at all but all core turbos for both are essentially "base clocks" until they start throttling as a default mode.
    This goes beyond software control unless you count UEFI/BIOS control as software control, which technically could be accounted as such but really isn't.

    Quote Originally Posted by Thunderball View Post
    Again, that doesnt need to be architecture or litography.
    In order to attain it with the 2nd leaked Single Core Turbo of 4,7GHz, Dual Core Turbo of 4,6GHz, Quad Core Turbo of 4,4GHz and All-Core Turbo of 4,3GHz... yes it does.
    In order to attain it with the original leaked SCT of 4,3GHz, DCT of 4,2GHz and ACT of 4,0GHz ... less so as speeds go down so increases power efficiency.
    TDP does not alter if power draw does not change.

    Quote Originally Posted by Thunderball View Post
    Well those Threadripper comments sounded totally unprofessional, but from a logical standpoint they are absolutely true.

    You can try slapping a 95W rated cooler on R7 1800X or 7700K right now, see how it goes. Gonna be the same result.
    Care to quickly change that latter statement?
    https://www.youtube.com/watch?v=ArJjI0F32rU

    I dare you to try that on a 7700K.

  15. #855
    Quote Originally Posted by Evildeffy View Post
    On official BIOSes it's plain impossible.
    iGPU still prevents BCLK overclocking and the silicon of the GPU is still powered even if the VRM are humming along at almost 0 load.
    You can still overclock an AMD APU (Kaveri/Godavari etc.) using base clocks if there's an external clock generator, still not possible on Intel's official BIOS and MicroCode.
    I guess I need to grab my multimeter and stab some mosfets if you can't do it yourself. And also overclock my 6700K on official BIOS through BCLK. You can overclock a K chip with BCLK if it has an external clock generator, non K overclocking is blocked through Intel Management Engine for Kaby Lake, you can still overclock Skylake chips on Z170 board with an external clock generator.

    You've been arguing that you cannot overclock through BCLK because of iGPU and now you're backing down?

    Quote Originally Posted by Evildeffy View Post
    You can "disable" the iGPU all you want, it's still powered and tied into the whole system bus, SuperMicro found a vulnerability to detach it from the bus clock and BCLK overclocking was reborn ... for a short while.
    SuperMicro found a way use the BCLK clock generator as a CPU BCLK generator, which also opened a possibility of non-Z chipset overclocking. To do it conventional way you just need an external clock generator, same way it works for Ryzen, but Ryzen has memory controller instead of an iGPU.

    Quote Originally Posted by Evildeffy View Post
    Go back to every HEDT platform and the CPUs prior (including even the Xeons, Core2 and Pentiums) which could ALL overclock with base clock changes, this is still possible because there's no iGPU tied into it, but w/e ... you clearly seem to change the subject to avoid talking about how wrong you were.
    Come on, try to set PCIe frequency to 120-140 (which was an average FSB overclock) on any LGA775 board, see how it goes (newsflash, you're going to kill it). Problem is not enough clock generators, not iGPU.

    Quote Originally Posted by Evildeffy View Post
    AMD's Vega is actually incredibly efficient on lower clock speeds so their APUs will likely be the same.
    AMD fanbois are still going strong I see.

    Quote Originally Posted by Evildeffy View Post
    Ryzen 7's All-Core boost is 3,7GHz and multi-threaded tasks still boost when all cores are manually OCed to 4GHz compared to stock, it's Single Core Boost (XFR) which is faster than such an OC.
    Intel's silicon/uArch/OC doesn't work that way at all but all core turbos for both are essentially "base clocks" until they start throttling as a default mode.
    This goes beyond software control unless you count UEFI/BIOS control as software control, which technically could be accounted as such but really isn't.
    Yes, Intel doesnt work like this, but there is nothing limiting Intel to implement something similar, at worst it's going to require a new chipset (which they are introducing anyway) and at best it's a just microcode change.

    Quote Originally Posted by Evildeffy View Post
    TDP does not alter if power draw does not change.
    So I guess you're saying that Vdroop, VRM heat dissipation and silicon having a better power efficiency under lower temperatures are all not a thing. Nice to know.

    Quote Originally Posted by Evildeffy View Post
    Care to quickly change that latter statement?
    https://www.youtube.com/watch?v=ArJjI0F32rU

    I dare you to try that on a 7700K.
    No overclock and open bench (same thing Paul did)? Sure, nothing to retract here. It's going to be fine in a closed case with decent airflow too.
    R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B

  16. #856
    Quote Originally Posted by Thunderball View Post
    VRM heat dissipation
    Wat? You do know how VRM's work right?

  17. #857
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    Quote Originally Posted by Thunderball View Post
    I guess I need to grab my multimeter and stab some mosfets if you can't do it yourself. And also overclock my 6700K on official BIOS through BCLK. You can overclock a K chip with BCLK if it has an external clock generator, non K overclocking is blocked through Intel Management Engine for Kaby Lake, you can still overclock Skylake chips on Z170 board with an external clock generator.

    You've been arguing that you cannot overclock through BCLK because of iGPU and now you're backing down?
    Show it to me that you can do it without the use of a modded BIOS on any chip that isn't a non-K Skylake chip.
    I will admit to assuming (assumption is the mother of all fuck-ups) they disabled it entirely for K chips as well, though I cannot really find any real "Yes it still works" points because of why you would do it when using multipliers is still quicker and safer.

    Multimeter still pointless are you're STILL failing to listen, the iGPU still prevents it because it's tied in with it, as well as the rest of the system if you go Haswell and below (PCIex etc.), the memory controller doesn't give a crap as that has multiple dividers and you can often get close to the original rated speed.

    The reason remains iGPU primarily and PCIex etc. in older gens, regardless if Kaby Lake's IME just flat out hates you if you try.

    Quote Originally Posted by Thunderball View Post
    SuperMicro found a way use the BCLK clock generator as a CPU BCLK generator, which also opened a possibility of non-Z chipset overclocking. To do it conventional way you just need an external clock generator, same way it works for Ryzen, but Ryzen has memory controller instead of an iGPU.
    And so does Intel, it's IMC is die-based as well...
    Intel incorporated it's first IMC on CPU die on Nehalem, AMD did so from the Athlon64 era.
    The technology (fun fact) is licensed to Intel from AMD.

    Quote Originally Posted by Thunderball View Post
    Come on, try to set PCIe frequency to 120-140 (which was an average FSB overclock) on any LGA775 board, see how it goes (newsflash, you're going to kill it). Problem is not enough clock generators, not iGPU.
    You never used the PCIe frequency on an LGA775 board to OC, you OCed the FSB which for the C2D/C2Q was actually very robust as you could push it to 400 on some chips/buses.
    My 990X and Rampage II Extreme can push 220MHz up from 133MHz without issue if I pour in the voltage.

    So tell me again why I should overclock my PCIe bus when it's separated from FSB?

    Quote Originally Posted by Thunderball View Post
    AMD fanbois are still going strong I see.
    Cool... you just called a lot of tech reviewers, including Gamer's Nexus, a fanboi... well done.
    The original Fury X was pretty efficient as well if it wasn't pushed to high heavens... the Pro Duo is proof of that.
    The APUs will follow that route over balls-to-the-wall overclocking.

    Quote Originally Posted by Thunderball View Post
    Yes, Intel doesnt work like this, but there is nothing limiting Intel to implement something similar, at worst it's going to require a new chipset (which they are introducing anyway) and at best it's a just microcode change.
    For what reason would they do so? They adhere to their own Turbo system... and they still adhere to the "system" I described prior.

    Quote Originally Posted by Thunderball View Post
    So I guess you're saying that Vdroop, VRM heat dissipation and silicon having a better power efficiency under lower temperatures are all not a thing. Nice to know.
    Vdroop doesn't account for ~60% power efficiency difference, it also doesn't lower TDP because it still uses X amount of power so the TDP rating still stands strong.
    VRM Heat dissipation doesn't alter CPU power usage, VRM react to the CPU not the other way around.

    And I didn't state power efficiency wouldn't improve, I stated that TDP wouldn't change if power draw does not.
    The efficiency will not alter the power draw enough to lower TDP to any meaningful value.

    Now stop twisting words when it is pretty clear what's written.

    Quote Originally Posted by Thunderball View Post
    No overclock and open bench (same thing Paul did)? Sure, nothing to retract here. It's going to be fine in a closed case with decent airflow too.
    Go ahead then... go cool a 7700K at stock settings with that passive 95W cooler, let's see what happens and if it throttles.
    If AMD can do it, so can Intel right? Even though the architectures are different etc.

    Especially considering the fact that that's a 65W TDP CPU for a rated 95W cooler... passively.
    And the Ryzen chips being soldered which Intel doesn't and it being 77 degrees under full stress, how will it react to a CPU that is rated 30W higher but is CONSIDERABLY hotter, you think it won't throttle?

    They aren't the same and they aren't comparable, understand that already.

    - - - Updated - - -

    Quote Originally Posted by mrgreenthump View Post
    Wat? You do know how VRM's work right?
    It should've been clear by now he has no idea how any of this works the moment he stated uArch of Ryzen and Kaby Lake being comparable.

  18. #858
    So, it looks like 8th gen U-chips are basically kabylake resfresh with 2 extra cores.

  19. #859
    The Lightbringer Evildeffy's Avatar
    15+ Year Old Account
    Join Date
    Jan 2009
    Location
    Nieuwegein, Netherlands
    Posts
    3,772
    https://videocardz.com/newz/intel-8t...of-300-chipset

    Now officially confirmed by Intel's own press release images.
    Z200 series boards are not compatible with upcoming Coffee Lake, earliest availability being mid-October.

    Still a short while away as well.

  20. #860
    I wonder if they will even state 8700K final clocks today or not


    still a while to wait then until 8700K game benchmarks

    at least its not 2018 or December



    - - - Updated - - -

    also - did they manage to fit 4/8 chips into the same 15W TDP previously held by 2/4 chips ? while having lower base clocks, but slightly higher boost clocks on 8650U/8550U ?

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •