Clevo Overclocker's Lounge

hmm. That C-Clip removal definitely needs the IHS to be at least 05-1mm protruding out of the bracket. It did wonders for me, @Prema, @Q937, however it barely did anything for @Mr. Fox mainly because, as his video showed, his bracket is actually inclined in a weird way.

 

I'm glad it worked out for you guys.

The idea is great but I have yet to see improvements without the C-Clip across a few systems.


Sent from my iPhone using Tapatalk

 

Well now we have the bracket lottery

 

Most of the functionality is now complete. The only thing missing is the temperature sensors (without them it just stays on max which I was testing with).

Time to make it presentable



This is a sheet of oven liner cut to be just smaller than the grill. I shall cut some nice looking holes in it but you can see where it will sit below:

 

@Meaker@Sager What fan brand/model you used?

 

Corsair magnetic levitation 140mm for the fans.

 

for that kind of money I would rather buy typhoons

 

The twin pack was actually not too bad, these are the version without LEDs of their own.

 

How is the Typhoons vs. Ek-Vardar fans for Liquid cooling?

 

I took the clamps out and it had zero effect since they were not even in the way to begin with. (For me that is)

 

Yeap. Same for @Mr. Fox.

When i did mine, the IHS was easily a mm higher than the bracket. Thats why it worked for me.

 

typhoons have been around longer

I trust nidec (typhoon oem) more than vardar/ek oem

 

But EKWB has been here for a long time. They are quite large on liquid cooling. The fans they push are set up for this type of use. I have both the 2200rpm and 3000rpm series fans. They are quite good.

 

Typhoons are actually considered one of the best 3k rpm pwm fans out there.

 

How do we fit it into the laptop tho

I thought we meant an actual typhoon and that it was a joke

 

I do not doubt this. Scythe is known for its CPU Coolers. As I know. But I haven't tried their fans.
Typhoons is discontinued fans + Scythe's newest 3000rpm fans is estimated with lower MTBF: 30,000 hours vs. EK-Vardar MTBF: 50,000 hours. And I don't find spec on what static pressure the new one deliver.

 

Seems like a pretty silly game, but so do all the others in this franchise. Not my cup of tea. Looks like a pretty weak system could handle it fine. This with everything maxed out, 1440p and no GPU overclock.

 

The engine is a little older now I think, how did the benchmark look?

 

Not any better than the others. Very cartoonish and unimpressive.

 

That's a shame.

I have tidied it up a bit. Will get another pack of washers for the 4 extra screws.

 

Damn, people still use flathead screws?????????? I only see those in wall plates for electrical for the most part.

 

They don't strip/mark as easily

 

It must be different in your country then. Because using the correct screwdriver with a phillips head will not strip and is also like 10 times faster.

 

Yes, if you have a perfect fit screwdriver it should not, for the flat head it matters less and IMO I prefer the look in this case

 

@Johnksss@iBUYPOWER @Mr. Fox @unclewebb

cpu throttled with the reason "ring: vr thermal alert" when single gpu (sli disabled) and cpu is loaded at the same time, ts limit reason "VR thermal" on core and cache


thoughts?

 

VRM's probably heating up.

 

http://forums.guru3d.com/showpost.php?p=5429350&postcount=354

 

Ok. I will go with personal opinion on that, just not the masses.

Can this be re created in real time. I mean while you watch everything and can see the actual point in time that the throttle happens?

 

Thanks.

 

yes

the cpu throttle start 20-25sec after 3dmark kicks in

 

but the throttle is almost immediate?

running standalone cpu stress, stays at 49x

 

So its happening only when the GPU's are in use simultaneously? And not with the CPU only stress? If its the latter, then VRM heating is probably not the issue.

 

Guys, c'mon we went through this same stupid nonsense with the DM-G and DM3, remember? I sure as hell do. Cancer EC needs @Prema to destroy their cah-cah code. Tornado F5 has a similar issue, but not as severe. Idiots design firmware now. Only a genius can fix their castrated firmware garbage. They're probably trying to stay under some retarded arbitrary power consumption limit and don't care if it works correctly.

 

cpu stress only works 49x without throttle


I used your advice on 2mm pad

 

Throttles to what exactly? 800 mhz or 2400 mhz or 4000 mhz or 4200 mhz?
Could be EC or could be you just don't have the amps set correctly.

What notebook are we talking about exactly?

 

I believe he is talking about the P870KM1.

 

Ah, thanks. @Mobius 1 usually has a bunch of notebook projects going on at the same time.

 

No problem. I think he is experiencing the same thing a lot of us did with the two predecessors to the KM1. If not the same, sounds pretty much like the same thing. Very frustrating, indeed. Finally has a real machine that's not a castrated turdbook only to have it crippled by cancer firmware. It might be wrong power settings, but Prema is already aware this model is cursed with severe CPU throttling on purpose by Clevo. Sure would be nice to know what kind of nonsense goes on in the heads of those goofy systems engineers at Clevo to make them think this kind of thing is somehow going to be OK.

 

km1 7700k 4.9ghz target static 1.19
drop down to a limit of around 70-75w if 1 gpu is loaded

 

Well, that's 100MHz higher that the artificially induced CPU TDP throttling the Tornado F5 has.

Run ThrottleStop and see if you get any Limit Reasons lights going off. If you're lucky you might be able to adjust them away, but I kind of doubt it. Definitely worth trying though.

Tornado F5 has a clock speed trigger that has nothing to do with power caps. Good up to 4.8GHz with no throttling whatsoever and at 4.9GHz instantly flips a magic switch. You're good for about 8 seconds at 4.9GHz or higher and then it drops to stock clock ratios. Sometimes I get lucky and can finish a 3DMark 11 Physics test before the switch is flipped, and with the right settings wPrime 32M will finish fast enough. Cinebench or anything longer than about 8 or 10 seconds and it will be running stock clocks until the load goes away.

Funny thing is, it can continue drawing over 100W at only 4.2GHz, so something is artificially reducing clock ratios. Behavior remains the same no matter what I set the amps or watts to... anywhere between BIOS defaults to maxed out, behavior never changes. Might be the same issue for the KM1.

 

It's because of how they "field test" the machine.

Going to need to see a real time example.
If it's in fact the EC, there is nothing you can do about it accept wait for them to fix it or return it.

 

LOL... Do you mean that tongue in cheek... as in how they don't field test them, or they do and use people that have no clue how to test anything? Hopefully, they're not pulling an Alienware on us and sending things out for field testing by people that do know what they are doing so they can more easily identify what needs to be sabotaged back in their laboratories. We (sadly) both know how that works, don't we Brother John. There are some pretty sick bastards out there with decision-making power.

 

No, they do test it, just not in the way we do. They can't seem to get past the old heads line of thinking while the new heads know they need to change.....But at the end of the day...it's who holds seniority that runs Clevo. And as much as everyone seems to think they are doing it intentionally as a whole is just wrong to think like that. And those are the facts. Basically you could say part of the company has been ready to move forward, but lack of seniority will always veto that.

 

Yeah, that's sad. And, I know it's true. Maybe we will get lucky and the old heads that are holding progress back will retire soon and the new heads will have free reign to do the right thing. Would be nicer if the old heads would either just stop being stupid or find something else to do with their lives.

 

Same arguments we have here are basically the same ones they have at Clevo. The difference is...You can actually lose your job, which most are not about to be doing anytime soon. Sort of why Sager will not be branching off on their own like all the other resellers can. They are locked in differently. They get to making too many waves and all hell will break loose over there as well. Do they ask for changes, sure? Yes, but it's more of a 1 time thing. 2015: Can we have over clocking? Clevo: Nope! 2016: Prema bios has been doing good things can we have a few more amps? Clevo: Nope! And ask one more time and find yourself on the chopping block. (Speculations of course) But you pretty much get the idea.

They do not need to be hex editing anything. They have the actual software and hardware to see and make the proper changes. So all that guess work is gone from the equation, but without proper authorization...It's irrelevant. It's not that hard to fix an EC if you have the right equipment, but if you have someone telling you you have to fit it in this envelope at this multiplier and at this speed and temp for this long....Well.... That does change things and we get what we have now. More problems.

 

vr thermal is the trigger

real time example?

 

Hmm, I haven't seen that one before. I wonder if that is an error in sensor readings, or a new magic throttle trick they discovered. If they set that VR Thermal Limit low enough, nothing is going to work right. Still might be able to tweak it away, but I am not optimistic. If nothing you can identify is getting hot and you're benching it on AC and still setting that error, it's probably accurate to say something is messed up.

The unlocked Prema BIOS has an option to disable Thermal Monitor, but there is nothing to clearly identify what thermal monitor. I wonder if that is the VR Thermal Monitor or something else. I've got it disabled.

 

That's not quite how the relationship works



Made some more progress on the stand I should be able to post up a video of the lighting and auto operation soon.

 

Might need to check if its making proper contact. Unless it the EC cancer that @Mr. Fox was talking about.
@Johnksss@iBUYPOWER did run 5+Ghz benchmarks using the stock BIOS IIRC.