*** Official Clevo P770ZM / Sager NP9772 and P770ZM-G / Sager NP9773 Owner's Lounge ***

I don't think manufacturers will RMA a GPU because of asic levels...

 

Not as such, no, but if the GPU runs abnormally hot under normal conditions then it should qualify for warranty replacement. That's assuming you haven't overclocked it because overclocking voids the warranty.

 

Indeed they are. This is also the time where I routinely kick myself for not choosing a faster shipping option. I always do this and never learn. Thankfully, I have some other toys arriving tomorrow to keep me busy until Tuesday when my laptop gets here. I am blocking off my work calendar that entire day (I work from home) so that nobody tries to get me into a meeting when all my attention will be on my new baby. Heh.

 

For anyone who's wondering, I have (almost) successfully installed Ubuntu 15.04 and nvidia drivers on P770zm. Correctly installing nvidia driver is a big pain, but it is doable. I still have to setup Killer wifi somehow though.

 

Please don´t open that box right next to your new precious


Dying Light on preset Best Quality runs fluidely. My HM with 7970M and 2960XM was sluggish on Medium:



Max temps CPU 75 C, GPU 65 C.
Min fps 40 (3 drops under 50), avg fps 69.
Recorded with ShadowPlay.
CPU-80/-100 mV, stock clocks. GPU stock.
Elevated on cooler, max fans.

 

So tomorrow's agenda: lap the new heatsink, verify contact and adjust if needed, relap cpu, apply fujipoly pads, compare temps to Kyle's, oc and benchmark.

 

Just in time for me to get my new card and run a new set of benchmarks

 

I'll be 970m standard, you'll have a 980m standard.

 

I know I know I'm just excited to have a new excuse to tinker haha.

 

Doh. Just got my laptop today and mine has a 69.9% rating. I sure as hell not going to RMA it, but it's still sad to know it's on the low end.

 

Yeah but remember ASIC is only an indicator, you can even OC far with low ASIC, it's all about silicon lottery.

Gesendet von meinem Nexus 5

 

Sort of. The limiting factor is heat dissipation. When you have sufficient dissipation, like with a good water cooling rig, the "lower" rated ASICs are actually better because they can run stably at higher voltages than ASICs with "higher" ratings. That's why water cooled cards typically have ASICs rated in the low to mid 60s.

Yeah, seems counter-intuitive at first but that's because "quality" in this context isn't the common usage.

 

I think he was just trying to state it's not an exact science. With my 64.2% asic score I was still able to OC more than some with much higher asic scores.

ASIC is a good indication/guideline that's about it

 

More gameplay. Witcher 2 on Ultra preset without UberSampling. Stock GPU and +200 MHz OC. i had to drop my CPU undervolt to -70 mV because of crashes.


Stock:
Max temps CPU 81 C, GPU 66 C.
Min fps 58, avg fps 86.


OC +200 MHz:
Max temps CPU 80 C, GPU 69 C.
Min fps 70, avg fps 93.

Quite a boost in fps by overclocking the gpu. Dying Light didn´t care, in fact I dropped 1 avg fps by overclocking.

 

I have a slightly higher ASIC with my 970M of 68.0% and my 970M OC's like a beast...

http://www.3dmark.com/fs/4349123.

@ajc9988 how does your 970M do in FS?

 

That is how mine was OC before, but now it's getting 1000 less. I'll figure it out later after I get the new heatsink finished!

Edit: what settings do you have?

 

Just got my new card installed and it reads 72.3% wheh much better than the 64.2% I had previously. Now let's see what this beast can do!!!

 

Now you're seeing one.
Fun fact: I'm not going to really play any game with this machine, it's meant to be a linux workhorse.

 

Wow you have a gem. Wanna trade j/k Please tell me you plan to OC that card!!! We have to know what it is capable of!!

 

Good to know I'm not hosed all around, but I doubt I'll ever watercool my laptop.

Why would higher leakage allow more stability at higher voltages?

 

Leakage is closely related to how much heat the ASIC generates. Keep this in mind.

Every ASIC is undervolted a little bit relative to the design ideal. The lower the "quality" rating, the more it is undervolted. The manufacturers do this to keep it within the TDP envelope of the design. Then everything else is adjusted so the whole package operates at spec.

Anyway, the "worse" GPU is undervolted more than the "better" GPU. This gives the "worse" GPU more room for increasing the voltage before the voltage exceeds the design specification and the whole package shorts itself out or shatters or something. More volts = more overclock. But! more volts = more leakage = more heat. And that brings it back around to water cooling.

I'm oversimplifying things, and I'll bet I got some detail wrong. This is all well above my rudimentary electronics background and the actual explanation makes my head spin.

 

Actually for overclocking a leaky transistor is also a fast transistor (unless it's just a dud transistor) so a more leaky transistor will have a higher top switching speed.

It comes down to gate lengths a lot.

 

Interesting note for anyone considering buying thermal pads from performance pc. I added some replacement pads to my cart as I plan to replace the pads on the rear of the GPU. I did not checkout and a few days later received an email for 5.5% off. It's not much but figured every little bit can help. I'll keep it there for a few more days and see if they move it to 10%

 

This assumes it even works, which is a fairly good assumption considering it passed qual somewhere in the process or they wouldn't have sold it. GPU's don't quite have as much bin'ing that CPU's do. I knew larger transistors (relative to the process) were faster, but had forgotten that leaky meant faster as well. Thanks for the backfill on knowledge. I knew all of this circuits business at some point in my past =p.

 

I think GPUs are binned far more than CPUs these days, intel don't really do very cut down versions of each of their chips, they have a dual core and a quad core which they slice little pieces off. However if you look at something like the GM204 it's shipped with vastly different configurations at different clocks.

 

Hi all,

I ordered this system last week based on your recommendations. I have the SM951 going into the m.2 slot and two 1TB Samsung Pro SSD's. I have asked the builder to put Windows 7 Pro on the SM951. I intend to put my Flight Simulator on the 1st 1TB SSD and all my photo scenery on the 2nd Samsung Pro.

I know just enough about m.2 slots to be dangerous
Here is what I am wondering. That 2nd m.2 slot is empty.

1. Is there any advantages or disadvantages to putting a drive in that slot? I see I can't put another SM951 in it so that tells me something must be different about it.

2. If I did put a drive in it, what could I do with it?

I guess it comes down to my thinking that as long as I'm stuffing this thing with everything else, why leave the slot empty?

Thanks for any suggestions and information. I really can't wait to fly the simulator with this beast.

Bobby

 

1. You would simply add another drive with more capacity for the system.

2. Anything you like. It's just another drive.

Have you considered raiding the two 1TB drives together to create a single 2TB volume for extra speed with large files?

 

Thank you Meaker,

I understand what you're saying about adding another drive. I was curious why my choices for the 2nd m.2 slot were not the same as the 1st slot. I'm starting to think it has something to do with the speed......something about pcie x4 and pcie x2. Anyway, I think I get it.

On the suggestion to raiding the 2 big drives.....evidently people who know how the flight simulator application works recommends putting the very large photo scenery files on a drive other than the one the simulator is on. There is a way in the simulator to "point" to the scenery files on another drive. This is why I got the 2 large drives. If I was using this machine for anything other than a dedicated flight simulator, I would raid it as you suggested.
I appreciate your help Meaker.......thank you.

Bobby

 

The reason the second m.2 drive had different options is it is unable to use 4pcie lanes. If you get an ssd using 4 lanes in the first m.2, you would want a sata m.2 for the second. Otherwise, the cheaper option is to raid two m.2 pcie, but it THEN limits it to 2pcie lanes per drive. Best performance is a single 4 lane nvme m.2 ssd. So hope that helps explain the m.2 options.

 

Thanks ajc9988,

That explains it perfectly. I have the SM951 going into slot 1 and so, as Meaker suggests, I want another drive for more space then I can drop a sata in that 2nd slot. OK......I got it now.

A lot changed in those 3 years since I got the P170EM

Thank you all.
Bobby

 

Your SM951 is a x4 PCIe SSD. It will use all 4 PCIe lanes available between the two slots. In the other M.2 slot you could put a M.2 SATA SSD for added storage.

Edit: ajc9988 covered it.

 

Thanks Bullrun, I appreciate the help.

Thank you to all those that helped me. I got "One Smart Nerd" to add the drive a few minutes ago. This is the final build. I'm in stage 2 with Mythlogic and waiting like it was 3 weeks before Christmas

Saved System Configuration
Mobile Chassis: MYTHLOGIC Phobos 1715 Chassis (Clevo P770ZM)
Mobile Display: 17.3" Full HD (1920x1080) IPS Matte LCD
Monitor Calibration: Free MYTH Professional Monitor Color Calibration
CPU: Intel Core i7-4790K 4.40GHz, 8MB Cache, Quad Core Processor Unlocked Devils Canyon (88W)
Thermal Compound: Free IC Diamond Thermal Compound
System Memory: 16GB (2 x 8GB) , PC3-14900, 1866MHz SODIMM
Video Adapter: nVIDIA GeForce GTX 980M 8GB GDDR5 nV GPU Performance Scaling
AC Adapter: 330W AC Adapter
m.2 Slots: Samsung 850 EVO 500GB m.2 SATA III 6GB/s Solid State Drive
m.2 Slots: Samsung 512GB SM951 m.2 PCIe x4 Solid Sate Drive
Hard Drive: Samsung 1TB 850 PRO Series SATA III 6Gb/s Solid State Drive
Hard Drive: Samsung 1TB 850 PRO Series SATA III 6Gb/s Solid State Drive

 

How's your Mythlogic experience so far and time frame (order to approx. ship)?

 

I agree,

That's one decked out setup.

I still think it may be better for you to install the main simulator on main PCIe drive and raid two 850 PRO's to have data files. I highly doubt that once fully loaded to memory, the fight simulator program won't need too much readings from the main drive...

 

When I got my P170EM 3 years ago from another vendor whom I shall not name, it was during the 7970M GPU debacle and I couldn't get any support from my vendor. I had 2 issues, one with the GPU and the other with the wireless card. Out of desperation, I put out a help post on the reseller forum and Mythlogic answered me right away. They stayed with me and helped me get parts and fix a competitors laptop. That impressed me and I promised them that my next laptop would be from them.

When I decided the P170EM wasn't cutting it anymore, I sent an email to Mythlogic in the middle of the night on a weekend and someone answered me back in a few minutes. That impressed me. So I was emailing them and posting some requests for information here and between Mythlogic customer service and the helpful folks here on NBR, I placed the order several days ago.

Today, I asked about the m.2 slot here....all of you were very helpful and then I got on the chat function in the Mythlogic site. they immediately answered my questions, set me up, sent a paypal invoice, and my system was updated in minutes. Since they assured me they build each system by hand and test it before it goes out the door, I will be satisfied whenever it gets here because I'd rather have a great job that took a little extra time than a faster shipment of a machine I have to RMA back because it wasn't tested. When I get it, I'll let you know how the whole process went but today I am happy with them

Bobby

 

Hi Raxel,

Ordinarily, I would go that route but there are a lot of addons to the main simulator that must go into the simulator sub-folders that would quickly fill the pcie drive. All the huge photoscenery files can go on a separate drive that I can make the simulator point to but some must read directly from the sim. This is why I'm putting Windows on the pcie drive, Flight simulators (and I have 2 simulators so both files together are pretty large) on the 1st 850 Pro and the scenery files (600 GB worth) on the 2nd 850 Pro.

You are right though, if that SM951 pcie drive came in 1TB capacity, I'd put the sims on them in a heartbeat for the read speeds. This is part of the reason for buying this beast. I was to the point of replacing the two 750GB spinners in the P170EM machine but I was also being bottlenecked with the CPU/GPU combo I had in it that I could not upgrade any further so when I saw the specs for this machine and read the first 100 pages of this thread.......well, this is why I am here today

Bobby

 

Yeah. Intel used to do it quite a lot more frequently. A fun example is the 66MHz 80486DX2 (remember that?). The failure rates in early production were so bad that Intel introduced two extra product lines to make use of some of the failures. First was the 80486SX which were the units that failed the FPU tests and shipped with the FPUs disabled. Second was the 80487 math co-processor for 80486SX which were the units that failed the CPU tests and had the CPUs disabled.

 

And here's hoping that you don't have to actually wait for three weeks before going into the build queue like I did with my Deimos.

 

I'm sure I read that 7970M thread.

I'm with you on the QA time.... take all you need to ship a well working unit... 72 hours testing, take 144 That is a huge benefit going with a builder. The personal care.

My only slight concern, as ratinox can attest, is "let me know", "keep me updated" if something comes up... simple communication. It actually, surprised me when I read about his experience and this http://forum.notebookreview.com/threads/reseller-that-offers-no-hdd-and-ram-options.774288/#post-9979921 It's great to hear it's all going smoothly for you. I look forward to the update.

 

Ooo, sounds like Malibal

Yes, I also agree with this. It is a fact that myth isn't as amazing as they were two years ago, but I am dead certain they're still head and shoulders above the others in terms of you getting a system working properly, and the available options with the system (except Eurocom; Eurocom just has everyone beat for options).

 

Still has me leaning to Myth. Though I want the amazing you got D2

And it seems Eurocom is better than they were two years ago regarding CS, from what I've read.

 

Oh they most certainly are.

Also, my machine's only merit right now is the screen. My CPU can't pass 3.8GHz and I need some modding and upgrading before I get to the 4790K + 980M OC levels of performance XD.

 

I would like to think Myth will do me right so I'll be positive. I'll keep you all informed. Meanwhile, I'm getting back up to speed reading all the threads I've missed since 2012.

Bobby

 

I'm positive you'll be fine.

 

Historically yes, these days not so much.

 

No one would buy a chip these days with a busted FPU (not that they are even separate anymore). These days it's mostly bin'ing. Occasionally designs will be manufactured with extra cache lines so that they can disable a few busted ones while still having the rated capacity. Cache's are huge and represent a large portion of most chips, and because of the way they work, it's irrelevant which entries are active.

 

Probably not failed FPUs, but if Intel were to have sufficient cache line failures in Skylake production that a budget line of i5 and i7 CPUs with, say, half the cache were warranted to make up for production shortfalls? They'd end up in budget priced PCs and people would buy those. They'd be big sellers with bulk purchasers, too.

 

Thanks for the feedback on this. I previously asked the same question in the Batman thread but never received a response. Good to hear that the 40x Processor Cache Ratio in XTU is normal.

 

I'm pretty sure that's how Celerons initially started. Chips with busted caches. Eventually they got better at disabling parts of the cache and Celerons became lower cache parts instead of zero cache.


More on topic: I ran Furmark for the first time on my P770M - 980m. GPU-Z said I capped out at 68*C and my PefrCap limiter was Power. But, I almost had 3x the frame rate of my 485m, and the clocks didn't look clipped. Is that "normal"? This is the first time using GPU-Z with the PerfCap field so I am not used to it. My previous machine would hit 90+*C easily. If this is running to full capacity at 68*C, thats fantastic.

 

@ajc9988 I'm curious how your mods worked out. I know you said everything is in and installed now. How is everything working out for ya?