*** Official Clevo P570WM | P570WM3 / Sager NP9570 Owners Lounge ***

if 16GB pairs can be found for about $120-130 now, then $60 for an 8GB stick seems like it's already here.

 

I just realized I don't need to find desktop users because this thread exists... XD.

I need some help from you lovely P570WM owners. I am asking you all because I absolutely detest primarily desktop-user forums who won't do anything properly. Who wants to be my guinea pig for testing about CPU usage?

I need you to have:
- Desktop CPU (ya'll have that cleared)
- Both stock speed (with turbo boost enabled) AND overclocked testing
- Sony Vegas 13, possibly GTA V or Dying Light. Best if you have cards like 980Ms in SLI to remove GPU bottlenecks, but not required (just easier).
- Second monitors is a very good idea, so that you can see information in real-time.
- Windows 8/8.1 and Windows 10 users. I am certain that Windows 7 doesn't have this issue, but if you wish to cross-check between 7 and 8/8.1/10 (if you have dual booting set up) that would be great.

What I need is for you to acquire Throttlestop 8 beta 2 and use it in monitoring mode, and also to check Windows Task Manager under the "performance" tab, at the same time.

I would like you to do your best to force CPU bottlenecks in GTA V and/or Dying Light, preferrably with the games in borderless windowed or regular windowed mode so you can snapshot the games running and task manager/throttlestop at the same time, something like this. I would also like if you can render any old video using Vegas 13 using x264 (Sony AVC codec works) in such a way that it uses 100% of your CPU (according to task manager, at least). I want to see what the discrepancies are between Task Manager and Throttlestop (the actual reading; it coincides with Playclaw and HWiNFO64 etc; unlike task manager).

I know using programs like TSBench or Firestrike's Physics test or XTU Stress test etc will actually use a full 100% of the CPU, but other programs (like GTA/DL/Vegas 13) can get limited by task manager's 100% level. I want to find out if the difference in 100% levels is a coefficient of max non-turbo speeds vs turbo boost/overclocking, or if it is a constant ~22-25% (as I have observed it to be for me). More samples are best!

If it is a constant ~23% number, then that means overclocking is a lot more efficient. If it does not affect your desktop chips very much, then it's something against mobile chips, which is even worse. If it is a scaling % based on base clock vs overclock, then that's even worse. So it's been itching at me and I want to know. Thanks in advance!

@Mr. Fox @sa7ina I think you two would be most likely able to easily help, so I tag you. Thanks!

 

I don't own Sony Vegas 13, GTA V or Dying Light.

 

I did at one point and it all worked perfectly fine.

 

Mmm well Dying Light and GTA and Vegas were more just programs I know to be affected. But I don't know anything else I can force a CPU bottleneck so easily for.

John have you never encountered any differences in the reported percentages? Not that I think many people other than me checked XD

 

Percentages as in what exactly?

 

In Windows 8 and Windows 10, Windows Task Manager's performance tab CPU utilization. The percentage util there is lower than the percentage util in programs like HWiNFO or Throttlestop's monitoring.

Some programs I've noticed get limited and won't use more CPU power once Task Manager says "100%" even if there's more CPU power. What I didn't know was whether or not it was tied to what Windows thinks the base clock is, or if it was a constant percentage difference. It's why I was asking people to check both stock and OCd for me just now, using desktop chips.

Unless you weren't replying to me when you said "it worked fine", in which case I misunderstood, sorry XD

 

I think your problems are temp related. When using more than 85 percent of the cpu power your chip needs to be be below a certain temp threshold (This is hard wired into every single chip and has nothing to do with what you see in temp monitoring software). If it isn't then you run into the behavior you describe.

And GTA V does use cpu power, but it uses far more gpu power in the end. Sony Vegas 13 uses far more cpu power than anything.

Best to run prime95 and find out. Since it allows for multi testing. (small/large/blend and custom)
Example:
GTA V=large
Vegas 13=small
wprime=blend

And they may bring about different results across different monitors or maybe they are the same.
I tested wprime right now and it says 100% on all 3 (TS/taskmanager/xtu)

Then find out if you program is stressing the cpu using the same method. Because without knowing that, it's just a guessing game of sorts.
Also, im using hexa cores and octa cores not quad cores....

 

I'm not sure if it's temp related. I often play BF4 and get CPU limited at 70c. I also livestream often and I use near (real) 100% quite a lot there, and that was never a problem even if I OC'd on a hot day and my CPU was at 90c for quite some time.

Yes, GTA V uses more GPU power, but I lowered some of the more demanding GPU settings (like grass and MSAA) to remove the GPU bottleneck most of the time. This is why I asked if people could try that; and why I said stronger GPUs would be better (easier to force a CPU bottleneck).

As for Vegas 13, here's a shot of task manager and throttlestop when I was rendering with it. Notice how WTM says ~99-100% but TS says I'm only at ~77% util? HWiNFO and playclaw etc would reflect the same thing as TS8.
But here's TSBench, which forces a full 100% like most benchmark programs. x264 streaming with OBS and x264 rendering with Handbrake also reaches this 100% limit, but Vegas 13 won't. It's why I specifically asked for non-benchmark programs to test, because benchmarks usually hit the real 100% limit.
Most games adopt the 100% WTM limit, and won't use more CPU power. Crysis 3 doesn't; I've seen that make my CPU hit 95%+ at times. Same with games that have heavy CPU load on a loading screen, like CoD: Black Ops 1 (sometimes my util will flash to straight 100% for a second or two when it is about ready to load a new map).

I know you're on a hexacore and octocore XD. Your specs in your sig often make me drool. I wish I could use a PC like that desktop you have XD.

 

Hummm, never tested any of them using windows 8 or 8.1 im afraid. Always windows 7 pro. Windows 8 and higher were known to have cpu "slowdowns" for no apparent reason at all. This was mention before it even went on sale. And it does it with everything (software and benching software) except for one benchmark program. And that was firestrike. And it gets worse with windows 10. Mr Fox posted some pictures on that.

For this to make sense, you would need to know exactly what say vegas 13 was capable of doing, then run it on windows 8 and check the completion time as well as the util.

I don't think anyone has found the answer either, just a lot of speculating over the years.

Edit:
Last time i did a file for a friend the whole 8.5 gig project took a total of 18 minutes from start to finish. Including burning the disk.

 

Yeah. I know Windows 7 DOESN'T have the issue for sure. Windows 8 and Windows 10 do, but I wanted to try and figure out if the util percentage difference was related to base clock, or if it worked different for desktop chips, or if it's a constant ~23%. But thanks for at least trying to help =D.

I use 8.1 primarily because certain things about it help me (with streaming as well, using monitor capture and such things) enough that I decided to stick with it over 7, but the CPU util issue is terrible. The thing is I want to know a lot more information before I try say... sending information to some websites, because apparently I'm the only person who ever noticed this. I mentioned it to unclewebb and he confirmed that it happens on Windows 10 as well, but he was unaware that some programs would stop taking CPU power as a result of it.

 

Well, when you find the answer please let us know. Because we have been asking for years now and no one seems to know why.

 

As soon as I find the root cause of the problem (be it CPU base clock dependency, or constant ~23% limit, or even hatred of laptop chips) I will definitely let you know, and provide detailed information to news websites about Windows' CPU limitations.

Us saying things mean nothing, but if there's irrefutable evidence about it and it makes a lot of big stories, MS will at least have to comment on it. Imagine all those overclockers or gamers with 4.5GHz i7 chips being told that their chips are still being seen as base clock for a bunch of games and programs, it'd mean OCing is worthless on the new OSes unless number chasing in benchmarks.

 

Well, i wouldn't go that far, but when running a 4.5 ghz overclock and someone with a 4.2 ghz overclock on windows 7 is better than yours....then best believe you are losing out somewhere.

 

@ Meaker, what are your thoughts on these? i thought intel mobile chipset dont support over 32gb of ram? which imo doesnt make a whole lot of sense when its direct connect to CPU rather than chipset. (although in the old days chipset does limit between 3 and 4 gb)

 

What part of a P570WM is a mobile chipset though

 

I've not noticed any CPU throttle in my machine at all under 10, odd. As for the ram some AMD systems and other chipsets can supprt larger 16GB dimms.

 

its not so technicially it should work but meaker mentioned anything before skylake will support more than 32gb. so anything thats 9x series chipset and lower. it would be difficult to think corsair would only allow 1 or 2 machine in laptop space supporting 4 dimms to have 64gb, but if its chipset then only p570wm can take it which make little sense, as corsair benefits nothing from it. p570wm owners are like 0.0001% of all owners lol if not less, i dont see how this will be chipset limited.

well desktop had 64gb with extended ATX until ddr4 came, otherwise it would stuck at 64gb. 128gb for desktop just got released with corsair's ddr4 ram.

 

Desktops could have two banks per channel though on x79 for a total of 8 dimms so the same capacity per slot.

 

@Meaker,
So you are saying that even though advertised as 64 GB ram capable, there is no current bios to support 16GB slots? Or are you saying that regardless the mobo will not be capable of using 16GB sticks?(for P570wm)

 

I believe the intel IGP controller issue is with more than 8GB per bank of memory.

So 8x8GB = fine, 4x16GB = not fine.

I believe that to be the case anyway.

 

It's more of a cpu slow down (Considered not running optimal). Easy enough to find out though.

 

Performance monitor in the task manager is not to be trusted. It can not even enter the correct clock frequency of the processor when it is oc. This applies in any case Win 8.1 and 10.

 

Well I know it's not to be trusted, but the problem is that it limits programs. That needs to be fixed.

 

Since my machine is going to be on show I have been watching it like a hawk on all my games and benchmarks and 10 has not been throttling anything.

 

I wasn't talking about a speed throttle of any kind, but rather a utilization issue. Could you check the screenshots? Or better yet, this long post I made on OCN (that place where those so-called enthusiasts seem to be incapable of batting an eyelash), where I explain in full detail what happens and what to look for?

 

its pretty tough what you are asking as a lot of these people dont have the same software/games as you do. I run intel burn test on my server 2012 (windows 8) and it cranked right up to 100%, and as accurate as throttlestop is, me and uncleweb have seen weird scenarios where it doesnt fully utilize CPU.

an example would be on xps m2010 taking a T7600G overclocking it past 2.8ghz would cause clock modulation/ chipset throttle by doing simple window 7 WEI test where as throttlestop's 32M test doesn't trigger the throttling and I have done this multiple times. although no software is perfect, throttlestop is one I trust the most even more than Hwinfo64, too bad unclewebb doesnt have gadget otherwise it'd rule imho.

back on topic, what im saying is win8/10 are probably more similar on fundamental level than win 7 but what do i know right? I just thinks win8/10 are similar but i have no proof whatsoever to back it up, however I do know windows 10 with the right IRST/eRST driver can give you a large performance boost on storage devices over windows 8 by a decent margin, its proven on an article recently done on tweaktown.

sometimes these softwares are purposely deigned not to take 100% of cpu usage as people like to do other things than just 1 game/program running and designer knows that. back in the old days i recalled windows 7 live windows movie maker dont take more than 60% of cpu usage on a crappy i7 4c/8t 2.0ghz CPU with turbo boost disabled.

its good you're looking into it though, as for enthusiast imho this is what we should check into to get highest numbers and every last bit of performance out of our machine. we are but a small fraction of a tiny percentage of pool of people out there who doesn't give a rat a** lol

 

Whatever metric is used to calculate the maximum usage is of little significance to me. If my FPS is the same what do I care?

100% CPU utilisation is a meaningless term if you think about all the units on the cpu as it is virtually impossible to get all units simultaneously executing at full usage.

 

Well this is the problem. I NEED those programs, because Linpack/TSBench/Cinebench/Passmark/3DMark Physics test/etc ALL bypass the limit. Only regular programs and some games seem to get limited by it. Games like Crysis 3 however do not. Games like BF4 and GTA V do.

I understand not everyone has the programs, and I'm not asking anyone to buy the programs, but it seems to be incredulously difficult to find someone who is willing to test. I've been asking around on reddit and twitch chats and all sorts of things and any time I get someone who'll go "yes, sure, I will test" they don't feel like using throttlestop or HWiNFO and rather use their own programs, then tell me "no difference" and won't follow my test guidelines. If I was rich or I worked at a place like Anandtech or Tom's Hardware I could just test on a bunch of different office machines, with varying specs, etc. I could copy my BF4 to an external and then then just populate all the PCs with it so we wouldn't have to download it a bunch of times. But unfortunately all I have is my laptop, and I can do no more testing than I already have. The only thing I'm thinking of is that I possibly could try lowering my turbo boost down from 35x and seeing if it changes things. But even if I did that, it still isn't enough data.

We already know how most people in the industry treat people with laptops.

 

in some way we can just conclude Crysis 3 is better coded for CPU, not sure lol

 

Yeah. I know. But I still wish to test the differences.

 

btw i thought u had m18x r2 or p570wm, not the new AW.

 

....
*copies signature*
Clevo P370SM3 (Mythlogic)| i7-4800MQ | 16GB DDR3L 1600MHz| GTX 780M SLI| 128GB Plextor M6M SSD | Samsung PM841 512GB SSD| 1TB 7200RPM Hitachi HDD | Intel 7260ac | 120Hz Matte screen
*end copy*

If I had a P570WM I would have been able to check with the desktop CPUs, instead of asking the users here to check.

 

bahhhh well guess i was blind

 

Hi
Don't have much time.
But i can help a little.
Can you please explain exactly what you need?
It looks like a list of ordinary tests and the net is full of this model's tests.
If you need anything (short) specific, do not afraid to ask.

 

Windows 8, 8.1 and 10 have an issue where certain programs (the four I listed I have confirmed to be affected) are limited by what Windows Task Manager's "performance" tab thinks is 100% CPU power.

This limit is different from the actual PC limit, and Throttlestop/HWiNFO64 show the correct CPU percentage (as does my Playclaw 5 program).

I want people to force a CPU bottleneck in the games I list, or render a video (especially using the codec I listed) in Sony Vegas, and compare the differences between Windows Task Manager's "100%" and Throttlestop's 100%. There should be a difference, and the programs I listed should be limited by that difference.

I also want at least a run with an overclock and without an overclock. I want to collect data to show that mobile CPUs are not the only ones affected, because it's hard to prove to media websites that laptops behave like desktops most of the time.

 

The only games i have now is Wolfenstein ET and the ones i develop.
And i do not edit videos so i do not have Sony Vegas.
Other Benches are fine?

 

Can you show some issue on your system, IE the lower utilisation listed AND a performance degredation to go with it at all. Once you have confirmed it on your own system then others can use that to see if they get the same behaviour.

 

But I already did, even though soon I'll be taking more examples.

My picture with Sony Vegas (even though I said I didn't show the rendering, and I need to find my installer for it because I had to refresh my PC some time ago and I can't seem to find it, or I'd run another test right now) showed that at 100% in Task Manager, it was not nearly 100% in throttlestop: HERE.
Then I showed that TSBench can easily bypass the Task Manager limit: HERE. You can see I still had vegas up in the background there, (also by the fact that the screenshot numbers are one after the other) as I ran TSBench right after the render finished.
I also showed how my GPU util was lower in GTA V while my CPU was maxed (but not according to throttlestop) HERE.
I can try to get one for Dying Light a bit later. EDIT: Did some Dying Light. GPU optimization got a bit better since I had last played, but I still got a CPU bottleneck instance where my i7 shouldn't be bottlenecking me: HERE

This is why many users who simply run AIDA64 or Cinebench or TSBench or whatever else never notice things, because they don't see it. Also, games like Crysis 3 aren't affected by the limits.

 

d2 Ultima heres my guess where window is flawed (in task manager)

in your picture, it says speed at 3.46ghz, however the graph is likely scaled to 2.7ghz (prior to any turbo) as you can see in the graph theres no indication its based on max turbo. if thats the case then it make sense that you are running at about 75% of cpu usage in throttlestop however in taskmanager graph showing ~2.7ghz which is at about 75% so maxed out.

now before jumping to responding, the top left graph simply showing speed but the actual graph itself is very likely scaled at 2.7ghz peak.

try this to find out, tone down your ghz with throttlestop to 1.3 or 1.4ghz and then run the software that will max cpu usage. throttlestop should show 100% on all cores due to low frequency however if graph is scaled to 2.7ghz and not change, it'll only show 50% load on graphs.

 

We know this makes TS report 100% correctly based on actual speed. But the problem lies in the fact that task manager will LIMIT program's performance based on whatever coefficient it seems to have. That's the real issue here. If it has some coefficient based on my base clock, I SHOULD NOT be limited by windows for that reason. That is a direct flaw in Microsoft's code.

 

test it first and we'll know, dont even know if thats true yet, my theory im just guessing. windows itself needs to run background files from time to time i do agree 25% seems drastic i'd say leave maybe 10% for windows and rest free to use.

 

Windows takes what it needs regardless. It's why having a bunch of (even idle) programs open and running benchmarks usually produces unfavourable results. Even at 100% it doesn't starve basic CPU needs for programs.

The problem is that Windows isn't taking up that last ~23%. It's just sitting there, freely, not doing anything. I.E. wasted power. Why bother giving me a 3.5GHz chip if most of my games use only 2.7GHz of it?

 

I'm not sure if it's what you where looking for. but it seems my GPUs are far weaker than my CPU and my game limited by their performance not the CPU.

 

Yeah... that's unfortunate. It's why I was asking for users to try and force a CPU bottleneck. Like, I added the Dying Light picture a while ago, but I could have turned down settings in the game and the FPS wouldn't have gone up. Same with GTA.

It might be hard to force a CPU bottleneck in the games I listed with say... a single 780M. It's easy for Vegas though. Interestingly enough, your task manager and throttlestop difference is very low. I would have to consider 2% within margin of error, unless it was consistent. But thanks anyway for trying =D.

 

thats not a window issue, more of a software issue not utilizing full cpu load. thinking of everything a windows fault just isnt right.

 

Windows Vista and 7 do not have the issue.
Windows 8, 8.1 and 10 have the issue.

I highly doubt that saying devs need to code around a windows issue that they don't know exists is the way to get around this.

 

Would you guys still advise buying a Clevo P570WM? Or is there a successor around the corner that I don't know of. In Europe the P570WM is very hard to find, I tried lots of resellers from the EU reseller thread. These are not eol are they?
I really want to buy this one, but don't know if I can still get it especially in Europe, and if I should wait for something even better that's possibly in the pipeline...

 

I don't know of any successor. Haswell-E draws a good bit more power than Ivy-E does; a 5960X would be near impossible in a notebook.

My assumption is that the P570WM is here to stay for a while, but users like Prema or Meaker may have far better insight into this.

 

or we can say win7 has a flaw not to show its true value lol, either way no real way to know as win7 had a ton of problem too. but overall i'd say win8/8.1 is much better due to USB BOT and UEFI and DEVSLP and bunch of other advantages over 7.

do u have any fps drop in win7 compare to win 8? did you only look at task manager and compare between two OS? i wouldnt trust anything windows throw at me really so i'd just time it myself if it was me. run a video encoding test or something where u have issue in win8 and now win7, time the two if they are same then no problem.