Sager (7280) Clevo (X7200) Power Supply Issues.

Tom's Hardware made review of the unit.
Is That A Notebook? MALIBAL's Six-Core, Dual-GPU, Speed Demon : Incomparable Performance?
They have similar findings to anandtech review. They run prim95 with 11 threads and futuremark on the both cards and tipped the power source. I am still trying to understand babyhemi's claims that it never happened to him.

 

I'm sure, Babyhemi didn't trip his PSU because for some reason during his tests only 1 GPU was fully utilized in Furmark. He also mentioned, that he's new to benching and could have made a mistake.
It seems that every system (980X+480X2) will trip PSU @ Prime(12 threads)+Furmark(Multi-GPU).
Let's see what Clevo has to say.

 

They will probably say that you are not supposed to be running prime and furmark at the same time. Overclocking on that machine only means you can do it, and Clevo never stated how much you can oc it.

 

^^^^Yes this is true. This was the first time I did Furmark and Prime 12 threads at the same time in a test on a system. What I did was before the test was to make sure SLI was fully enabled, some how and I do not know how, only one GPU showed up in the final result after an hour. Maybe it deselected to prevent a PSU trip, ...not sure...

 

make a youtube video of the test.

done deal

 

At this moment it seems that it can be used only the over-clocked cpu at full load or the 2 cards at full load, but not in the same time. No point to buy all of them. If there would be no problems, I would be happy with a hexacore at 3GHZ and 1x480m.
The other point from both tom and anandtech is that the 2x480m are equivalent in performance with 2x GTS 450 cards. While the 480m costs 800 $ each the GTS 450 costs 130 $ each.

 

That would be a bad idea. Some users reported PSU tripping during gaming. It won't take long till more people reproduce the issue.

 

i'll have my ram this afternoon and will start to play..

i said it before.. but when i got my 8800m gtx.. the gtx desktop variant was 200.00 and the laptop was 700.00. and it wasn't comparable to a 8800 gtx. it was maybe** to a downclocked 8800 GS. and the gs was prob a 150.00 card at the time

 

I wonder if the clevo power utility, is used how much performance is it lost, though it can only be checked if someone uses a custom power source.
Also if the power problem is solved, how will in general behave the motherboard in long time when been under so much stress.
Some coworker claimed that my D901C would not last too long because of the level of load, but the x7200 needs more power than the D901C so this statement would extent to it as well.

 

my d901c lasted a long time.. and i had the cpu pin-modded and the gfx card overvolted

 

Without being argumentative that's correct, you shouldn't be running 12 threads of Primes and furmark at the same time. It's not even close to real world usage. Running Prime95 as you all know (assuming blend test) runs the CPU to virtually 100% and also the RAM pretty hard. If you then add 98% load on every transistor on both GPUs then, although you can easilly argue that the system should be able to cope with this, it's way above what the hardware is supposed to run sensibly.

If people are experiencing PSU issues when just running games, just running heavy encoding software or heavy CUDA apps then it would suggest an issue with that specific piece of hardware.

Basically what I'm saying is running Furmark by itself is ok but it's still a GPU burner, running it with any other intensive app is ott. Prime95 and 3DMark looping one of the GPU tests (for example Canyon Run at 1920 which uses a decent ammount of GPU RAM) is a good test of stability, running Prime95 and Furmark is not a good test - be it on a laptop or desktop setup.

 

You have a good point. But what is it then reasonable load?
90% cpu + 100% 2x gpu?
80% cpu + 100% 2x gpu?
70% cpu + 100% 2x gpu?
60% cpu + 100% 2x gpu?
50% cpu + 100% 2x gpu?
40% cpu + 100% 2x gpu?

 

Exactly. There is no real definition of unreasonable load. While we all agree that running prime and furmark at the same time are extreme and don't represent much in real life experiences, a computer should be able to handle maximum load. The PSU tripping issue is unrealistic (as in does not happen in normal coding, gaming, etc), but that does not justify it.

 

Partially true, IMHO. This specific combination of tests is a standard procedure for most serious reviewers and being used for years on all laptops/desktops (even netbooks) to determine the power stability and temp. issues.
The test is totally necessary. What if tomorrow GTA 5/6/7 appears and utilizes the CPU 80-100% on all cores? Together with high graphics demands it will surely trip the PSU. Will you say - "This game isn't compatible with x7200"?
The problem is not in the GPU's, for that matter, even if M5870 CF is used, the tripping would (very likely) still occur. It is in the desktop i7's and the amount of juice they consume under full load. There's no easy way to fix the problem w/o crippling the performance.
A better PSU is needed, capable of supplying up to 500W+, maybe a bridge of 2 300W PSU's. If Clevo chooses to go the AW route of BIOS tricks - I can't even imagine the level negative reaction all over the place.

 

That is in part the problem - what is reasonable load?

On the GTA 5/6/7 argument, Tomshardware ran Primes and Crysis simultaneously and there were no problems with the PSU tripping out.

"The good news that followed this test was that no combination of real-world programs and games was able to create the same phenomena, with Crysis and Prime95 pulling a maximum of 372 W from the wall"

They didn't mention anything else about the PSU nor did they conclude that it was a negative point.

In my experience different reviewers tend to take very different approaches. For example in the UK the premier PC modding magasine (won't mention specific names) pushes desktop PCs way more than any other review magasine or website. They test with Prime blend test and loop Canyon Run for 24 hours and if it fails this it doesn't get reviewed. This is for stability and max power consumption and is much stricter than any other media reviewers I have worked with in 6 years. They even decided not to use Deep Freeze test because it had a habit of killing GPUs after 24h testing. On the other hand you have websites and magasines which class full CPU load as the Cinebench benchmark and also report max CPU temps from that rather than Primes.

Anyways, I'm going to see how far I can push an X7200 over the next couple of days if I have time and see where the "breaking" point is, though we've not had any problems and nothing has been reported and I know that several have gone out to companies that do serious CUDA work.

 

good sir, to end all worries, is the psu of this unit (x7200) future proof to future next gen games(like crysis 2 or next gen games)? i mean, when new games come out,i want to use my future x7200 for gaming and i dont want it to trip when playing next gen games.

btw-i dont oc or benchmark too much so i should be fine,yes?

 

Good points, Neil.
Looking forward to your findings! If you could do some video encoding + games (GTA, Metro, etc) or pro applications + gaming/rendering, - would be great.
I'll probably join the testing/benching/OC'ing euphoria soon

 

I do not think it is known. That is the question I guess. What is the reasonable load?

 

i have a question about the benchmark link in toms hardware

will the i950 + dual 480m gtx have the same performance with the i980 + dual 480m gtx on the crysis and call of duty benchmark( on the high and very high details)?

 

^^ Should be very close, IMHO.

 

That would be great! My notebook appears to be at my local UPS Office, but they will only deliver it on Saturday since Friday is a Public Holiday..

 

i cant understand. according to this link Review Deviltech HellMachine DTX (Clevo X7200) Notebook - Notebookcheck.net Reviews

it states crysis on very high no AA but 30.9 fps

but this link Benchmark Results: Call Of Duty: Modern Warfare 2 And Crysis : Is That A Notebook? MALIBAL's Six-Core, Dual-GPU, Speed Demon

crysis with very high 4 xAA 32.9 fps

both have dual 480m gtx sli. whats the catch? does the cpu make the difference?

 

The CPU. The first one uses an i7-960 CPU the second one uses an i7-980 CPU.

 

Hi

Regarding the "SLI" Notebook power supply addapter for the clevo x7200 I've read about, had it been done yet, and has any one tried it?

Sorry if this has been asked before.

 

does the cpu make a huge difference in gaming most especially on crysis? i thought it relies on gpu only?

 

Don't know if you have seen this, a later update to the AnandTech review: Update, 10/16/2010: One of the readers commented that another site was reporting issues with power draw exceeding specs and causing the power brick to shut down when running games, specifically Mafia 2. I spent most of yesterday playing Mafia 2 on the X7200 with no issues, at maximum detail with and without PhysX enabled (four hours on each setup). Power draw at the wall reached a maximum sustained value of 335W, with a few instances where it briefly hit 340W. That should still be under the maximum 300W rating, assuming 80% efficiency, but anything more (i.e. a minor overclock of the CPU) would push the system over the edge.

 

I would say 2 frames is a very minor difference, and in fact well within the error margins. But sure, some games uses processor powers more than others - Crysis is demanding on both processor and gpu. GTA is another example of another, even more PCU dependent game.
Really, I would expect to see a couple of frames difference between a i7 950 or 960 and a 980X.

 

the thing is,the first link shows with 4xAA used and has 32 fps while the other link shows 30.1 fps with no AA. does it mean that w/o AA on the i7980,it will boost more fps than 32?

 

We had one person, Speedy that had an issue while playing Mafia 2. If I remember correctly he had already tripped the PSU with Prime/Furmark, so it could have been a faulty PSU issue.

 

GTO -
Remember, this is not 100% identical machines; there's differences in the hardware. Memory, SSD vs. spin disk, differences in installed software and drivers maybe, differences in the processes, foreground and background, running ... and processor. You can't really directly compare them, unless you took one of them and installed a different processor in it, and then compared. And there is always margins of error.

 

Yes, and I suspect this is the case AnandTech is referring to.

 

Since speedy is the only one so far that had issues when gaming, it's not an issue that needs to be considered seriously. Not yet..

 

...untill someone plays Mafia while reformating a video file in Premiere and using 3ds Studio to render a huge 3d scene at the same time...

 

I OC my 980x to 4ghz, I ran it with prime95 small fft's, all 12 threads, and furmark on my single 480m, it reached a whopping 411w before my psu tripped, not bad for a 300w psu

 

Dexgo

I'm waiting to see your benchmark tests..... My order got pushed till november 16th cause of backorder 480s. Debating canceling it and waiting for ATI announcements at the end of the month. All my MMO friends are calling me a moron for buying the melting laptop.

 

Sounds like your friends are just jealous. Lol

 

Naw I'm kinda iffy about buying it... unless I can see some great benchmarking figures about the Heat and Watt consumption. I owned 2 Clevo D900's in the past and they would burn thru mobo's and graphic cards faster then I could get replacements. I don't have time to own another lemon. Having the arrival time pushed up 3 weeks is giving me time to think about my terrible purchase as of late. I really don't care how fast this thing is, I care if its gonna run stable and that is not gonna burn my fingers like my D900's did

 

From what I have gathered reading these threads a couple of times is that when it is not OCed, it is stable like a rock and heat is not really an issue. That's why I got the 480s. It's a strong card at stock and mild OC settings. It might not be efficent per watt per watt like some other cards but it is all relative.

The Bugatti Veyron gets like 2mpg per gallon at top speed but I don't think you will see people complain about that.

 

i haven't had much time to do any real benches bud. i had to order a mini pci-e card for wifi. so i can do updates and download drivers.. there are no eth cables in my house right now.. and i don't feel like pulling cable 100' and splicing ends... not that that's hard.. but i ordered a atheros mini from vancouver for 15 bux.. it will be here tomorrow or this coming monday.

I did install crysis. with stock drivers.. there were some vsync issues for me and some stuttering..

I need at minimum a month with this thing before I get ustu her quirks... and feel her out. ebb and flo

 

This is what I was looking for. Could you please run the same test (12 threads prime and one gpu furmark) but without the cpu over-clocked? In the not so close future I am planning to buy the x7200 with hexacore and single gpu and I am curious if at stock speedsfor cpu and full load the adapter is enough.

 

Hopefully this is a driver related glitch. Keep us posted when you have all the drivers in place. I'd also recommend running DPCLat Checker while gaming to see the actual latency.

 

did some tweaking. vsync only shows now with the cut scenes.. all i have tested is crysis.. my benches are bone stock. stock drivers that were included in the package from clevo. I also don't have that much ram.. I have 2x 4gb on order.. i had to pay mucho cash to pick up a 2gb stick locally.. just so i didn't have to wait...

I ran crysis bench @ very high no AA dx10 x64 and got avg 26fps steady. the normal fps reader read 30.. but avg was indeed 26.xx fps.. i let it run 4 times per test. first run was avg 16.. but we all know the first bench is a dry run.

I also ran it with single GPU. it was about almost halved (the fps) so sli is doing it's job.

that's all i got right now.. my time is limited cuz of work and such.. and having no wifi card.. when the card comes.. and the weekend.. I will get into gear.

take care

dex


p.s when i get the net on the beast. i will include some thumbnail screenshots.. i just don't have the time right now transfer to this pc to post.

 

Ok. The original statement I made also had a 480m gpu OC of
Core 530mhz
Shader 1060mhz
Mem 1350mhz
So thats a 4ghz 980x OC plus the above gave 411w before psu trip.

Now stock everything, P95 12 threads small fft's, plus furmark, only 315w, the psu can handle that all day.

I even have throttlestop on which enables the cpu turbo feature so cpu was at 3467mhz, 133x26. The x7200 when maxed doesnt use turbo, cpu clocks stay at 3333mhz 133x25.

How clevo every decided on 300w psu for and sli system I will never know, plus they encouraged overclocking as a feature too, with no power to sustain it lol.

 

I have this feeling that XTU was done but they didn't have that extra time to fully test it out , clevo just need this Systems on the consumer market and see what's the respond before any revision for improvment.

 

Thanks. I guess I will be safe at 3 GHZ and no GPU overclocking. I will go with a hexacore if it is not over 600 euros , otherwise with a quad core when they release the 32 nm versions. Some of the xeons hexacores are rated as 80 or 95 W TDP so I have now hopes that my setup will be stable. There was another userr who put a similar question, though I am not sure of he wanted one or two graphic cards.

 

will upgrading the ram affect the performance on gaming?

 

2-> 4Gb - yes, 6->>12gb - no

If you do buncha VT, rendering, and keep several games minimized - you'll be low on RAM

 

dex,can u try crysis on all high with max res? w hat fps do you get?

 

im planning to pick up an SSD tomorrow for my system.
i need help on installing windows and bring all the drivers to my new SSD,
is there a guide or something i could follow for the x7200/Sager NP7280?
im afraid i might screw up something..

 

Don't try to clone an ssd or merge partitions from other drive i learn my lesson is easier and faster start from scratch just put all the drivers on a folder of your secondary hdd and install windows on the ssd then install all the drivers that will save you a lot of headaches