X9100 Results on 7815u!

You can trust ThrottleStop temperature and MHz monitoring. When you unlock and adjust the multiplier higher, some programs like Everest are not prepared for that change and will start reporting the wrong MHz. I don't believe that CPU-Z has any problem after you unlock a Core 2 CPU.

An X9100 has two cores and ThrottleStop should be reporting temperature data from each core. Post a screen shot of ThrottleStop if it does not show two temperatures. When fully loaded with Prime95 Small FFTs, the temperature of both cores should be very similar if you have accurate sensors on your CPU. Intel used some poor quality sensors on their 45nm Core 2 CPUs, especially on some of their ES CPUs.

No temperature sensor is 100% accurate. Your core temperature is not that important as long as your CPU is stable.

 

Ok, thanks Unclewebb.

I meant that throttlestop shows one temp at system tray, minimized. It is now showing 34 C on my P7801. Everest is showing 42 C for CPU, 33 C for CPU core 1 and 33 C for CPU core 2.

I have googled it and I understood that CPU for everest is Tcase and for CPU core 1 and core 2 is Tjunction.

Ok, I understand that stability is more important than the actual temp. ie, if your system is stable is because your temp is good, isnt' it?

But I guess temps could be related to the life of a CPU, evem a system itself, so I supose running a colder system will bring you a longer components life.

One question, unclewebb, which temp trhotlestop is showing at system tray?

Compared to everest, I can assume it is Tjuncion. Is Tjuncion more important than Tcase?

Once again, thanks for all support and state-of-art throttlestop.

 

ThrottleStop reports CPU core temperature. The system tray will show the hottest value of all cores. If core 0 is 35C and core 1 is 40C then the system tray will show 40C. Tcase is a meaningless temperature to monitor. The only accurate way to measure the true Tcase temperature is to cut a groove in the top of the heatspreader on top of the CPU and mount a thermocouple to the center of the heatspreader.

That's why Intel include core sensors which are designed for user monitoring. Thermal throttling and thermal shutdown are both controlled by the core temperature so there is no point in monitoring any other sensor unless your core sensors are broken.

The core sensors are not very accurate at low temperatures but usually work great at higher temperatures near the throttling and calibration point.

 

ok, no more doubts.

Really Thanks.

 

Using now 3,7Ghz with 1,325 V.

Played FSX for 40 min, super stable. Processors not always at 100%

Temp max CPU 84 C; GPU 73 C.

Idles at 32 C CPU and 47 C GPU.

Room temp 24 C.

No notebook cooling, only raised the back a little.

Increased a couple of FPS.

Should I try evem more?

 

I still have a p8400 and that seems amazing.

 

That sounds good. You can try for more just don't get too crazy with the voltage.

 

Tried to multiplier 14,5 but throtlestop doesn't change to the new clock.

I guess the limite is 14.

Fair enough, 3,7 is fantastic already.

I'm gonna try overclock the GPU, I don't know how to do it yet, but I'll do some research.

If I can gain more 5 FPS, I'll pay the beer to everybody lol.

Erik, thanks a lot, people here in NBR are superb.

 

sup guys: How is your gateway P7815u running with the x9100 CPU, it is running really hot??

 

Still running nicely: idle temps ~30 °C, under load @ 3.5 GHz ~60-70, but with Prime95 ofc had the higher temps. Overall the chip has been running very well past few months.

 

I'm good on 3.99GHz on 15x @ 1.450V. Stable when gaming but prone to auto-shutdown when parts of the game (DoW2/Ninja Blade/DoW: SS/Prototype/Darksiders) become intense.

Idle temps in the region of 49C. Under load: 98C/(>102C)

doesnt happen too often though. My fallback compromise is 3.6GHz @ 1.300V

 

Interesting and good to know. May if you manage to lower the temps it could become more stable.

How you can go further the 14,5 multiplier?

My throttlestop does not allow me to pass 14,5.

 

That good to hear i am maybe updating my friends gateway with x9100. If i cant Find one malle t9800 or t9900

 

Quite impressive, I had an X7800 in my 6860 about 1.5 years ago, I oc'ed it to around 3.2 Ghz was in the high 10's in 3dmark, considering getting a X9100 for my sig rig.

 

I was able to pull off some descent overclocking
3.6Ghz at 1.25v
3.4Ghz at 1.2v
and on the low end
2.8Ghz at 1v!

It runs pretty well, I can stay at 3.6Ghz all through Red Faction G. which is a crazy insensive game.

 

you're completely stable at 3.6Ghz @ 1.25v without BSOD?! I experience BSOD on that OC setting which is why i stay with 1.3v or 1.288v.

 

I can run there but orthose errors right out. I have to run much higher for full stability. While I could run notrmally lower I like the security of knowing I can flood the CPU with 100% stability.............

 

1.25v for 3.6ghz is absolutely crazy, is it an E0 stepping x9100? My C0 x9100 runs well at 1.2125v/3.33ghz, but to take it to the next notch(3.46ghz), I got to pump the voltage up to 1.3v, which is really stupid. Interesting thing is once I get passed the dreaded 3.46ghz, I only need another 0.3v to make it run stable at 3.6hz, I tested with orthos at 1.33v/3.6ghz, so far it hasn't produced any errors. Temp hit a maximum of 89 C during the torturing test. I guess it's not too bad, huh?


Edit:

3.6Ghz at 1.25v is amazing, but testing stability with a game is a no go, I can stay at 1.2625v/3.46ghz stable in games, ps2/wii emulation(very CPU intensive), but orthos always gives me errors until I increase the voltage to 1.3v! If you experience any errors while running orthos or wprime, it doesn't matter if your games run fine without crashing, it's going to cause problems in the long run, some rather big problems. I asked unclewebb specifically about this.

 

I have been stable at 3.6Ghz at 1.25v while testing with Wprime and while playing Red Faction Guerrilla and it has been fine. I haven't had any odd errors or anything. I had a blue screen or two when trying to find a stable setting and the laptop getting too hot.

 

I could not even touch wprime dual thread at 1.25 as it would error right out. You have a very good sample of a x9100 then as most will not touch stability at that low a vcore.

 

Ok so I noticed I may not be so lucky. I have Throttlestop set to 1.25v however CPU-z and Wprime are both reporting 1.4v I don't know which is right here....

 

Mine seems to work ok..................

 

Are you using the latest version of CPUz or an older one?

 

Infiniteone: When you have a problem, post a combined screen shot of both ThrottleStop and CPU-Z so we can have a look. You need to have ThrottleStop enabled and you need to have a check mark in the Set Multiplier box so both the FID and VID requests will be sent to the CPU. I added a VID window to the latest beta version so it will be easier to see what values you are actually getting.

ThrottleStop 2.90 beta 6
http://www.mediafire.com/?jdal8a26utytlum

 

The reason I posted a full screen shot, so they could compare versions and options...........

 

Good thinking TANWare.

Good you try the ThrottleStop beta version I posted? I just want to see how it looks. For some reason version 2.89 is getting the right side chopped off a little with the theme you're using. I like your theme so if I have to, I'll widen TS a few pixels so it doesn't get chopped. Do you have a link for that theme so I can do some testing? My version of Visual C++ dates back to the Windows 98 era so once in a while there are a few bugs. I'm surprised my tools work at all on Windows 7.

 

Hi, this is using windowsblinds. It gets a bit funky at timed with some apps but overall it is much less boring than jus the stock windows interface. I'll DL it and let you know.......

 

Ok so I was using CPU-z 1.5 now that I got the latest version I can confirm I am stable at 3.6Ghz-1.25v

 

Hi Guys,
I'm new to this thread but I was reading all the posts and felt enthused about buying the x9100. So, I did. I too have the p-7805u. I have a few general questions and problems that you guys can hopefully help me out with.

1st: When running prime95, do you choose 1 or 2 torture test threads to run? Two seems to be the default. When doing so, my temperature cycles between about 93C and 63C. Is this normal? I am running the test completely stock. I attached the screenshots as it is cycling.

2nd: Does anyone know why when I try to change my multiplier and voltage to what others have seem to be stable, I get a crash in prime95 almost immediately?

In addition, I used Arctic Silver and NC2000 for cooling.

Thanks in Advance.

 

stevebibu: Your CPU is running too hot and it looks like it is thermal throttling. The Intel throttling is not this severe and is designed to rapidly cycle on and off so the CPU speed is maintained a lot better than what you are seeing.

It looks like your laptop has something else built into it that is in control of this throttling. Basically what happens is your CPU heats up, triggers the throttling, that reduces the CPU multiplier down to 6.0 and it also reduces the VID voltage so your CPU cools down. After it cools down, the multiplier and VID both go back to their maximum settings, your CPU heats back up and triggers throttling again. It cycles like this over and over again.

Your CPU heatsink is either not tight enough or not adequate for this CPU. If you are having trouble keeping your CPU cool at the default MHz settings then you're not going to have any success trying to overclock it which only creates more heat. I'd pull your laptop apart and make sure your heatsink is clean and tight to the CPU. You might have to try and re-engineer it if it is poorly designed.

Hopefully someone else with this laptop model and heatsink can give you some temperature numbers to compare to.

 

Hi Unclewebb, thanks for the prompt and insightful response. I should expose the copper fins on the Northbridge by scratching off the black coating, right? I know that the cpu and heatsink are making good contacts as shown by the thermal compound spread when attaching the heatsink. I may have put too much arctic silver on the cpu initially; however, I have since reapplied an appropriate amount(too much is just as bad as having none, correct?). Could this have permanently ruined the cpu? The integrity of the cpu looks 100%. Again, thanks in advance.

 

this sounds very worrying. please elaborate what you are actually doing to your heatsink and/or northbridge chipset.

 

Here is a picture of the cpu and thermal compound to show you the amounts. It looks alright. Do you know if there are any sort of diagnostic tests to determine the physical integrity of the cpu. I might have gotten a bad one. As far as Mod's, I have not performed any on my computer. I tested the CPU after I put it in. I am going to put my p8400 back in and perform some tests with prime95 right now.

 

here is the pic, sorry.

 

To follow up on my previous post, I reinstalled my p8400 and have been running prime95. The temperature plateaued at 63C. So, I do not think it is an issue with the heatsink or my thermal compound application. I have emailed shirleyfu and will hopefully be able to exchange it for another one.

 

are you also applying artic silver to the northbridge? If so, that's a big no no.

AS5 is far too thin a compund to bridge the gap between northbridge and heatsink. Probably explains your temps. You either need to copper shim if you insist on using AS5 or switch over to a thicker TIM compound like ICD7.

I idle at 27C on stock 3.06GHz @ 1.137v with ICD7 applied on both the X9100and northbridge.

 

The Northbridge is the part of the heatsink that sits over the chipset, correct? I removed the thick paste sitting over the chipset and replaced it with the arctic silver. That said, can that explain the high temperatures of the x9100 CPU? I currently have my p8400 in my laptop with arctic silver over both. Prime95 yields a temp of 62C. Does this seems high for stock p8400? I still have the thick paste off of my chipset that I can put back on and test it out with my x9100, If you think that will help. In either case, I order ICD7 the other day and should get it quite soon.

 

the northbridge is the chipset. It has nothing to do with the heatsink. Presumably, the "thick paste" you are referring to was the stock thermal pad used by Gateway - does it look a small squishy blue square.

Like i said previously, replacing the stock thermal pad/thick paste with AS5 over the northbirdge chipset was a bad move. It's too thin to bridge the gap between heatsink and chipset. As such, there is a very small air gap which results in bad thermal conductivity to the heatsink which means high temps.

ICD7 is a superior upgrade to the stock thermal pad or you could copper shim the chipset by purchasing a 14mm (L) x 14mm (W) x 0.9mm (H) copper plate and sandwiching it in 2 layers of AS5. Place this inbetween chipset and heatsink to bridge the gap.

in any case, ICD7 is by far the best way to go. Never had a p8400 so cant really comment on what its temps should be like.

 

Thanks, Hello Moto. I'll hold off on exchanging the x9100 until I try your suggestions.

 

Hi Hello Moto, I CANT BELIEVE IT!!! You were right, the thermal pad on the chipset solved the problem. Considering the integrity of the pad is not 100%, i was still hovering around 72C with the x9100. That sure beast the 98C with the arctic silver over the chipset. With the ICD7 which should be arriving Monday, I will hopefully bring that down some more. On an aside, when speedfan or throttle stop register the temperatures, is one temp that of the chipset and the other of the processor?

 

glad to hear your problem is (partially) solved. The temps shown by throttlestop refer to the 2 cores of the X9100; the chipset does not have a thermal sensor and therefore has no temperature data to report.

when you apply the ICD7, make sure to place a pea-sized blob of the stuff (approx. 5mm in diameter) on the centre of the X9100 and chipset. Then compress down with the heatsink before securing in place with the mounting screws.

 

Thanks a bunch Hello Motto!! I replaced the arctic silver and thermal pad with ICD7. prime95 is causing a temperature of 61C. That is about 13C cooler than what I saw with the other configuration. This stuff is absolutely amazing and is an absolute must for anyone considering getting this CPU.

Final Modification: Sitting on NC2000, ICD7 on chipset and x9100

 

Has anyone had any luck using E8435? $89.99 on fleabay right now, has the same specs as a T9900.

 

The E8435 might idle at a high temp, don't think it supports EIST.

 

thank you, I may just go for the x9100, good price, good speed.

 

hey everyone: How is the x9100 going for you stevebibu?I am wondering since I got a gateway fx 7805u and I am thinking on updating the CPU sometime. I am not sure if I should go with the t9900 or the x9100.

 

Go with the X9100. I upgraded recently and the overclocking has done wonders in games, especially Fallout, Oblivion and Bad Company 2.

 

Your gateway is up again?

 

I don't believe your clock settings. There is no way that the Gateway cooling system can handle running a game at 4.0Ghz with a X9100 chip.

 

Not being a gamer I can't say every game can, I know mine with lower CPU usage could. IE 3DMark06 would run with no issue other than the CPU stress section. On the CPU section it would borderline overheat so if it were extended for further time there would have been an issue.

Some games are harder on the CPU so there would be an issue. As far as Games the one I have, Dirt 2, would run at 4.0 GHz but I saw no benefit over 3.2 GHz so instead of burning the CPU I just did not bother..........