Nvidia clockblock: vBIOS (unblocked in 353.00)

I hear ya and like I said, I don't disagree. I'll just leave it at my last post. Good stuff though. I also wanted to say good bloody example! That compliment of course has nothing to do with the fact that I just placed an order for that exact same notebook series, haha. You must be looking over my shoulder *looks back*. Oh and please don't use that fact to justify your side of the discussion, because I know exactly how you can do that . Ain't "rolling with it", just had to compromise a bit for my needs .

 

The only throttle table I don't mind is a heat-induced one. My 280M used to run clear past 105c at times when things were too hot and depending on the game etc. I'm perfectly fine with a throttle point somewhere around say 92-95c, especially with cards that can handle 100-105c. But any other kind of throttle? No thanks. That's far enough. Take it and go away with it.

 

Yeah temp throttle is separate and comes on top of that...

I found the sweet spot to be at 92c for my Maxwell Mods. Up from stock 87c, as most of Clevo systems only spin the FANs to 100% at 89c, so temp throttle @ 87c is a performance waste in Clevo land unless the user keeps the FANs manually on full blast all the time...

NVIDIA's Mobile Maxwell default system shutdown target (via system EC controller) is 94c sustained for more than 30s, so we can't go any higher either.

 

It would be nice if Clevo would fix the fan table so if it kicks on, it stays on like it did with 880M. The way it is now, it kicks on at 90C and back off at 88C so it's on, off, on, off all session unless you force max fans.

 

The best thing to do is to allow tiers of fan speeds, or to simply let the hotkey software choose when to use 100% instead of 80% max. Even though I have no problems setting my fans to 100% all day, I rather keep them on auto if I need to due to dust buildup etc (the air in my house can get quite dusty) and let them kick into 100% if they pass say... 72c and kick down below 60c

 

Even the Hotkey software with the custom fan profiles that max at 80% don't have any effect... Lol

 

Even on newer clevos? I know the SM-A one was more of a test and the original hotkey app looks nothing like it.

 

I have the latest SM-A Hotkey and the latest EC (I'm running Clevo stock 308 BIOS and EC). Adjusting the custom profile does not appear to have any effect at all. I don't know about other models.

 

Hang on now, do auto fans ramp up to 100% (fn+1 speed) when the GPU passes 89C? I thought it never went above 80%?

 

That is correct. They ramp up to 80% and then drop back to what I would imagine is the 20% that they go to as soon as a 3D app is launched. There is no middle... Its really obnoxious. The 880M hit 87C and it went to 80% and stayed there even if the card dropped down to the 70s. I'd much rather have that back than the forced 100% only option. The way the stock machine is set up, the fans never kick up (because the card throttles the core at 87C).

 

You focus at infinity whilst using the Rift due to the in-built lenses. Completely different kettle of fish to what you tried years ago.

That's why using the Rift for long sessions is potentially going to be better for your eyes than using a monitor.

I've even read of the Rift correcting peoples lack of stereoscopic vision.

I suggest you give VR another go, you may be pleasantly surprised.

 

Yeah... no. I've seen this spiel before and it's never panned out so I'm not getting my hopes up.

 

This conversation actually fits better here:

I did not reinstall Windows 7 or 8, but I did change the BIOS to Legacy (ran a Windows repair so the OS would boot using BCD instead of Boot Manager). I also tried with the stock vBIOS and the publicly available @svl7 vBIOS and the all of the behaviors are exactly the same. I think there is something in the system BIOS causing issues with NVIDIA cancer drivers, and whatever that issue is seems very similar to the throttling malfunction with 980M. The thing that gets me is the "WHY NOW?" part of the deal. I mean, everything has worked flawlessly for the M18xR2 until the release of Maxwell, and now that they are focusing all of their energy on drivers for Maxwell GPUs everything turns to crap. I don't know what to think of this. Nothing has changed with the end-of-life M18xR2 Über-beast. So, I can't blame Dell/Alienware for something they had absolutely nothing to do with. The only thing that has changed is products from the stupid retards at NVIDIA. It may be only a few Clevos that have Kepler GPUs that do not malfunction that are the rare exception here rather than my Alienware, because people rocking Kepler desktops are having a lot of grief as well. And, things are not all sunshine and roses for Clevo owners either... some of them are having issues. This does not make me feel better, but miserly does love company.

This may also explain why there was no Maxwell refresh with a Ranger 2.0 or Viking 2.0. If something NVIDIA has done with Maxwell firmware and drivers refuses to play nice with InsydeH20 BIOS, that's a problem.

Not only is this behavior with my 780M cards having severe throttling with all drivers after 345.20 similar to 980M behavior in the M18xR2, it's also similar to 880M behavior in all machines, which has never really been fixed for anyone. So yeah, nasty finger points to NVIDIA... not going to blame Dell/Alienware for the incompetence of the Jolly Green Giant. He needs to start taking Viagra or something, because he's losing his mojo.

 

So, i overclocked my GTX 850m with nvidia Inspector. anyone knows how to save the clocks and apply at startup ?

 

Which Windows OS are you on and is UAC enabled?

 

Win 8.1.

i finally used MSI afterburner and set apply at startup. but now i have some problems

I Play GTA V .. i decreased the laptop ram to 8 gb for upgrade and now i am getting many drops in GPU usage

the game uses 6 Gb of ram could that be the problem causing the GPU drops?

70 C is my max temp for CPU
65 C for GPU

 

nVidia is becoming the new AMD lol

No but seriously I feel your pain, and even on the desktop side Maxwell drivers seem less polished compared to the Kepler days so I hear ya.

 

353.06 has just been released! Let's put it to the test!

 

You're a bit late to the party.

 

Whoops, didn't notice that. Received the alarm from Geforce Experience just some minutes ago.

 

Use Google translate This is the future... It does not look bright. Amd begin now and block overclocking of VRAM on the new desktop graphics card. What next??? All overclocking of graphics card for gaming laptops will surely be removed gradually. A frightening future for us oc enthusiasts . Written by Sweden's largest online magazine about computers and hardware.
http://www.sweclockers.com/nyhet/20734-amd-sparrar-overklockning-av-hbm-minnet-med-radeon-fiji

 

Yep, so that's applicable to AMD's new HMB related architecture. The article sites the reason for OC of the VRAM to be blocked is due to it being a new technology and that future drivers may unlock the overclocking function of the VRAM. I'm OK with that if there are technical reasons behind that - no point in wrecking your card, maybe they need to be cautious with the new HMB VRAM until they know what it can handle for the long term. You can still OC the core, and HMB does offer massive memory bandwidth anyway.

 

That is the new direction I did not like... Closing for overclocking of VRAM is not a good idea anyway. What next??? It is the customer who must be allowed to decide over their own purchased hardware... Amd and Nvidia is going to do all possible devilry for us who want to overclock our own graphics cards in laptops in the future. This is just the beginning of the end...
http://www.legitreviews.com/amd-fiji-arrives-radeon-r9-fury-x-details_166515#OpQZsfoAmLt9ajjt.99

 

Well, Google translate doesn't work for that web site for some reason, and I do not understand what is written there. But, I don't give a rat's shiny hiney what their rationale is, if it won't overclock because they've done something to bolster their controlling god complex it's a pile of trash and needs to be regarded as such. They can used their rationale as a suppository. If they don't know what long-term durability is, they should not be selling it, and besides that, it's just none of their stinking business what end-users do with it.

 

I'm going to play devil's advocate here, because if there is a problem and lots of customers accidentally fry their cards then that's a class action lawsuit begging to happen. nVidia recently lost the class action suit over GTX 970 RAM performance; they don't need another one.

 

Well, in that case don't use the public for the R&D test-bed. Hold off on releasing it, test it, overclock the crap out of it, and then if it's durable enough you can sell it in a year or two. Sell it as a locked down piece of doo-doo and nobody will have a clue whether it's really any good or not.

You know, all this "progress" trend that essentially levels the playing field and makes everyone exactly the same and removes end user control is Satanic. The entire concept is from the pits of hell and I feel sad for the folks that are eager to let these pick-pocket companies replace their brains and rule their souls. So, I guess playing devil's advocate here is fitting.

That's awesome if NVIDIA lost that class action suit. They deserved to lose it for false advertisement and lying to customers, and selling defective products. They deserved to have their pants sued off for using drivers to manipulate overall system behavior as well. The moment their drivers do something other than drive a display adapter they are trespassers.

 

http://www.legitreviews.com/amd-fiji-arrives-radeon-r9-fury-x-details_166515#OpQZsfoAmLt9ajjt.99
AMD feels that the memory technology is too new and there is more than enough bandwidth... LoL

 

Yeah, well... as usual, AMD can stick what they feel where the sun don't shine. We can all see where their opinions about what it means to be awesome landed them and it's not a very good place. It seems to be pretty rare that they ever get anything right. Their lackluster performance in both CPU and graphics are the major reasons NVIDIA and Intel are both producing so much garbage. Intel and NVIDIA don't even have to try hard right now, and their mistakes are accomplishments in comparison to AMD "achievements" LOL.

So, AMD thinks having an SoC model that combines vRAM and processor into a single chip is a good idea? These folks are more retarded than I had ever imagined. Man, things are so screwed up in the tech world right now it's not even funny. They abandoned us for consoles and now they are trying to inject us with their console cancer approach to engineering... nice!

 

Amd had the opportunity to do something now. But they waste the opportunity as usual... I look with horror what comes when the updated GPU and processors in laptops being launched 2015/16. Even more frightening perhaps autum 2016/17.

 

Well, from what I read the only reason they are able to have such a High Bandwidth card was due to the very fact they did include the VRAM along with the GPU core. I read that because it's so close it enables them to have such a wide memory bus - something along those lines.

 

Speaking for myself, I agree.

But that's not how mainstream consumer electronics manufacturing works. There are no design fixes. There are no refreshes. Mass market is the beta test. The ODMs make notes of problems and add them to the QC checklists for the next model line. That's what we got when let the MBAs take over from the Engineers.

I don't think that's the way it should be but it's what we're stuck with for the foreseeable future.

 

Unless enough of us bludgeon them with bad press and force them to change to avoid going out of business. If we draw blood, they will react. If we don't, they will continue to rejoice in their own unrecognized failures.

If they are going to market an enthusiast product, it needs to be worthy of being purchased by enthusiasts. This is not worthy, and the least we can do is highlight their failure(s) to help them get better. This piece of crap is instantly disqualified by virtue of its over-the-top retarded SoC design. The Kool-Aid drinkers think that going to an SoC model is nice, but we expect them to be stupid and so far they are at least meeting that expectation, LOL.

That sort sounds like a good excuse on paper, or nice spin to put on it for media distribution. Ooops, uh-oh, look... all out of Kool-Aid.

 

They've put the VRAM right next to the GPU core for a reason though. Likely cost/performance reasons. Maybe they could have designed a product with the VRAM away from the GPU core, but perhaps to get the same high bandwidth then the product might have cost more to make, hence also costing more to buy.

 

Yeah, of course... SoC garbage is the trend. That's the reason, not because they're awesome. Having a reason doesn't excuse bad decisions. SoC is a self-serving strategy to sell more trash that cost less to make. If they are wanting us to accept this as a mainstream hardware solution, that's fine. Added costs are expected by enthusiasts. Low cost and high performance are like oil and water. They don't mix and if you try to force them to it turns to a slimy sludge. Enthusiasts pay big bucks for the good stuff. If it can't be overclocked, it fails to meet the litmus test of "good stuff" LOL.

 

Well @Mr. Fox you might as well get out of GPUs altogether seeing as HBM is the direction the industry is moving as a whole and Nvidia Pascal will have it next year.

As for me, I welcome HBM with open arms as we will finally get past the 256-bit limitation that traditional DDR/GDDR has had on mobile GPUs for years.

 

Yep, but still it's bandwidth is higher than the previous generation - that's a good thing. I may have also read that they were reaching the limits of what GDDR5 can do anyway, you can't just go on increasing the clock speed I don't think, and creating massive memory buses to compensate is expensive I think.

 

Probably will... or, maybe just enjoy using old hardware to make the kids ashamed of drinking the Kool-Aid with their fancy new emasculated trash, LOL. We can already get that concept of garbage from AMD in a console. No point in paying extra for it in discrete hardware that is crippled. There's no doubt in my mind they developed this for a next console and are now simply trying to make a fast buck spinning it off as something special for PC.

 

There will be a cooling problem for VRAM with this new design. Then you can not overclock the VRAM because of heat..

 

Perhaps so!

 

Yeah, it's got liquid cooling for a reason, and that reason is not because they are so desperately trying to appeal to enthusiasts. It will need to be crippled even further to avoid creating a throttle-fest or fire hazard in a notebook. When you rev an engine too hard it's going to break. Going faster without spending more money, something is going to be broken. But, they like to refer to that as "efficient" because it sounds nicer than crippled.

 

Increasing memory frequency is possible seeing as desktop GDDR5 is up to 7 GHz (8 GHz with overclocking) while mobile GDDR5 has never gone above 5 GHz (6 GHZ with overclocking), but not practical because of heat and power considerations. Going wider than 256-bit is physically impossible in the current MXM form factor without HBM, and the power & PCB space savings could potentially allow us to have bigger dies and more robust power circuitry on mobile cards. Not to mention, the vast increase in memory bandwidth from going super-wide instead of super-fast will be a boon for high resolution displays, antialiasing, VR, and the more detailed scenes enabled by low-level APIs.

There is also an air-cooled Fury with identical specs to the Fury X.

 

It will be interesting to see how the air-cooled version behaves in real-world usage.

 

That's cool, good to know, I'm more of fan than I was before of HBM now you've discussed the limitations of the current VRAM designs in relation to notebooks. (I think we were talking about the desktop versions of the card before). I have no problem with AMD's latest HBM card, I think it's a pretty good effort from them to hit equal (and in some cases better) Maxwell in 28nm form. HBM makes sense to me, more so now you've shown the advantages for laptops with HBM. Lets see what they can come up with for the next gen AMD (and NVidia) cards!

 

Hey guys, is there any good tutorial to overclock the 970m??? i have the ms-16f4 and i'm looking to see if i can push this card more, can someone give me some light or hints?!?!

 

Did you really just open a year-old thread to ask how to overclock your laptop?

 

Hey guys, how's the weather on Hawaii?

 

I agree with the others. Check out @jaybee83's statements recently for testing for 24/7 stable overclock (recent postings) and his OP in the P750zm thread. Lmao!

Sent from my SM-G900P using Tapatalk

 

Yeah there was no reason to bump this thread of all threads. Closed.