*** Official Clevo P65xSA/SE/SG / Sager NP8650/51/52 Owner´s Lounge ***

May I ask you why it is unacceptable? In all my tests I noticed absolutely no performance loss. (Including bf4 online multiplayer)

 

Some people do work which uses 100% CPU and having a faster one saves lots of time.

Also I will be getting a 980M which I will OC and want to make sure I'm never CPU-bound.

 

Well it depnds on how you load it, 3.6ghz is a 1 core turbo so you can chuck that out the window for multi core loads, 3.4ghz is the 4 core clock. Then they have specced a 47W limit so that is the most you can get out of the arch for that speed under stress conditions.

Games will stay clocked up. Turbo was always about making the most of a TDP window rather than shipping a 2.6ghz CPU that runs under all conditions.

 

Just played about an hour of bf4 yesterday, CPU was around 60 degrees, as was the GPU. Game was on ultra running @ 60fps. To say I'm satisfied is an understatement... Now if only BIOS 306 from prema would be released.... grrr

 

Quoted these from earlier pages to show my previous tests for CPU throttling.

 

That looks normal for haswell.

 

I am assuming doomed with those throttle limits in place? Any way to improve them?

 

Thats a picture of the stress test. I thought I was supposed to test during the benchmark? The problems I have are not visable during the XTU stress test (see attached pic). In reality it makes no difference for me since, in your words, it throttles like crazy during regular usage (gaming, work).

How come I cannot access the turbo boost power time window nor the processor current limit (is this due to me not using Prema mod)? The option is simply grayed out or not there for me (see pic).

 

Yes, you will need Prema Mod to do so.

 

How did you stop your power throttle...sliders in XTU are not doing anything for me on stock bios.

When running XTU stress test with -80mV undervolt there is no throttle but once I start up PRIME95 power throttle goes to 100% and CPU downclocks to 3.0Ghz

Are you running Prema BIOS and that's why you can adjust turbo power limit?

Are you using XTU to adjust it or T.S.?

 

@LoneSyndal

Deleted due to the extreme ineptitude of my eyes

 

Its a -50.

 

My bad...for some reason the first time I opened the pic I didn't see the minus.

 

@LoneSyndal

Did you experience the same problems that I am currently having before you flashed prema mod and tinkered with the settings? Our laptops have basically the same specs (except I have the 4710HQ) but your pictures of running XTU shows completely different "results".

How does your CPU look while gaming (or game benchmarks)? And how does it look of you run XTU with everything on default/stock?

Thank you for your help!

 

Adding my few cents to the bowl, to weight LoneSyndal a bit more: same settings, same results, TDP capped at 58W and clock at 3.4 under P95 and TS. First P95 run is without TS, second is with TS, same XTU settings for both. (i7 4720)

 

@jeanjackstyle

How does your CPU perform at stock settings? Do you experience fluctuations on processor frequency while gaming (assuming that you do game on the laptop )?
Thank you!

 

Hello! I'm new to the forums but was wondering what people's view was on the 970m vs the 980m? My priorities are keeping cool and portability while playing some older games that are not hugely system intensive (my favorites are mostly older games like WoW, Portal, Half Life 2), while still being able to play new games occasionally (this is lower priority). I don't mind paying the $250 extra.

Thank you!

 

they are roughly the same size machine so if you don't mind paying the 250 difference for the better card just get the SG instead of the SE

 

Second quote was forced max clocks if possible, but would throttle under load.

 

I just completed a 9 hour Prime95 blend test with my CPU undervolted by 90mv without error.

The strange part came when I stopped Prime95. 3 minutes later, at idle, the laptop blue screened.

What the heck is a likely explanation for that?

 

Guys, I see you talking about the throttling and i'm a bit worried. Can anyone take a look at my benchmark?

The cooling paste is arctic mx-4 extreme thermal conductivity compound, maybe the applied it wrong or didn't even apply it? 98 degrees is not good I think.

The first run if with fans on automatic and the second run with the fans on there maximum power. Btw everything is on stock settings.

Thanks

 

It means your undervolt isn't stable. Drop it. I had same situation on my batman. -100 and -90 bs'ed right away. Thought -85 and -80 were stable but were not. I am -75mv and haven't crashed in days. And yes, I primed, xtu'ed and others. It's just not stable.

 

Is it not a little odd the the undervolt worked in Prime95 for 9 hours then it only choked at idle?

 

Not to interupt this conversation, but i have the same issue but much less problematic. My cou goes up adn down to at around 65% and the up again. The cpu does never go above 80 degrees and its hovering around 3-3,2 ghz.

Is it worth to RMA it and will they "accept" it as a defective unit?
Btw im using the latest official BIOS.

Thank you

 

It's not strange at all. In fact, it's quite typical. Because voltage drops down to almost nothing at idle, undervolting is more likely to cause BSOD at idle than under load.

 

I just tested undervolting myy CPU (a 2.6Ghz i4720) by -50mV in XTU (I have the Prema Mod Bios), the results slightly surprised me, as the CPU temps were higher, not lower, after the undervolt - I thought undervolting would result in lower CPU temps?

These are the benchmark test results:

Stock:

Score: 788
Highest CPU Temp: 77 C

50mV Undervolt:

Score: 805
Highest CPU Temp: 86 C

Is this to be expected? I experienced no thermal throttling with either config during the benchmarks, not in a 5 minute CPU stress test

 

Your benchmark looks just like mine except with much higher temps. So if there is something wrong with my laptop I would guess that there is something wrong with yours aswell (even without regard to the high temps).

So another one who has the same issue. My question to the ones who have more "normal" behaving processors remains; did you experience this kind of throttling/unstability before tinkering with XTU and modded bios? All three of us are on stock bios I would presume (me and IceBounds at least).

 

It looks like your clock speed was increased as a result of your undervolt. I suspect that's why your temp went up.

 

So I just started Prime95 to check if the graphs would look as stable as yours @LoneSyndal and @jeanjackstyle and I guess it kind of does.

So my laptop is not faulty then? It is just that this machine has a hard time every time you want to utilize CPU and GPU simultaneously (weird considering this is a gaming rig after all ).

Has anyone had better results using a 240W or 330W PSU?

Still find it strange that the laptop struggles as much as it does. My previous P150SM with 780m never had any issues like this despite it too being used with a 180W PSU.

 

For the power thing, have to wait until a new 240W arrives. My last one was a dud.

There may be other limitations that we cannot see such as the motherboard support.

I think I hit TDP throttle in an earlier quote if that says anything.

 

Really? I didn't mean to, I just meant to undervolt the CPU only! This is what my current setup in XTU looks like - what do I need to change to just get the -50mV undervolt? BTW once set in XTU, will this "stick" when I reboot next into Windows or do I need to change it in the Bios somehow?:

 

It happens naturally due to how Turbo Boost works. Notice how your performance also went up after the undervolt.

 

This is all stuff I'm learning right now too.

You undervolted your CPU which, obviously, means it draws less power and runs cooler, which in turn means that the CPU has additional clockspeed and thermal headroom before it hits either the TDP, current, or thermal maximum and throttles itself.

Right now I'm struggling to understand the current limit, which is what causes my CPU to throttle itself most of the time. I've downloaded Throttlestop and I think I see where I can fiddle with the current limit, but I don't know how safe that is to do.

 

Exactly, good job

 

Does anyone have issues with XTU not loading your profile when you start it? i can't even load my saved profiles and i have to manually set my undervolt every reboot.

 

I guess I need to revise my statement above. Currently I'm testing a -60mv underclock. What I'm seeing is that it depends on what program I use to put load on the CPU. In Prime95 I'm seeing current limit throttling almost exclusively. With Intel XTU's stress test and encoding using Handbrake, I see power limit throttling almost exclusively.

If anyone wants to comment on that and help me understand this better, I'd sure appreciate it.

 

But mine ran 9C hotter (if faster) when undervolted, which is not really what I wanted. Should I return it to stock voltage for lower temps?

 

You could underclock while keeping the undervolt, thereby getting lower temps with the same performance as stock voltage

 

BTW just ran it again at -50mV and got exactly the same results for the XTU benchmark - 805 score, 86 C max temp. This time the computer has been sat cold for 20 minutes, but the result is the same.

 

Guys I wanna make a point.
XTU Bench seems to throttle a lot on mobile chips. I don't know why, but it happens. XTU stress and AIDA64 or something like encoding a shadowplay video at "veryfast" in Handbrake will be the real tells on if your CPU can't handle a lot of load.

 

Which XTU setting "underclocks" and what would be a good initial setting to try?

Would be interesting to see if I can get the same score as on stock (788), but lower temps (77 C)

 

I'm getting no thermal or power limit throttling, but lots of limit throttling - I don't know what the latter means, though

 

In XTU, under Manual Settings and then under Core, you can change (lower) the multiplier sliders to underclock.

 

Neither me. All I know is I usually get knocked down to 3.3GHz in XTU Bench at some points regardless of what I do with my system. The only thing I haven't tried is static voltage at 1.192v (the highest I've seen my system draw at stock adaptive voltage) and see if that works, but I'm not interested enough to do so, since XTU bench is much more related to good RAM than it is to a good CPU, and I don't have good RAM (yet).

 

I'll add my results to the mix:

On stock voltage, running XTU Stress Test, I hit 100% CPU utilization, clocks hold steady at 3.29GHz, and I have zero throttling; temps sit in the low 90s. But when I run the CPU Benchmark, I hit 100% CPU utilization, get "spiky" current limit throttling, and my clocks fall to a pretty steady 2.99GHz.

With a -50mV undervolt (I'm going to play around with it some more to find how low I can go), the behavior in the Stress Test is the same, except the temps are about 5-7 degrees lower. In the CPU Benchmark, I still get current limit throttling, but less of it, and my clocks are hitting 3.09GHz.

What would be causing all of the current throttling in the Benchmark, that doesn't show up in the Stress Test?

 

Thanks for the quick reply and helpful responses D2! So if I'm hitting the power limit throttle, is it accurate that I have the following options:

1) Raise the TDP bottleneck (can this be done in XTU) but risk overdrawing power from the 180W PSU? At which point, maybe purchase a larger PSU which I believe Lone is going to test?
2) Pull back on the amount I'm undervolting at the cost of more heat
3) Leave it be and accept the consequences, which I'm guessing will manifest itself as stutters/spikes when gaming? This isn't an entirely out of the question option for me as putting all 4 cores under full load is not likely and when it does, the power throttling was very intermittent

--
I'm also starting to see minor Current Limit Throttling but also very intermittent. How is that different from Power Limit Throttling? What is it telling me and what knobs should one turn to adjust for it? Or is it even a big deal at all?

 

Yeah you're right, on the Stress test everything is perfect, but in the benchmark it's all spikey.

 

Power = Watts
Current =Ampers
Voltage= Volts

Power = current x voltage

Watts =Amps x Volts

 

Yeah, but if the Stress Test manages to push the CPU to 100% utilization and hits clocks of 3.3GHz, where is the extra current draw coming from in the Benchmark that causes the clocks to throttle to 3GHz?

 

When I dropped the multiplier on 4 CPUs to 33x (from 34x) on -50mV I got 806 benchmark score (highest yet) and temps of only 77 C!

I'm not sure whether I should leave it like that and/or drop the multiplier on any of the other core numbers (1, 2, 3) though.