*Official* NBR Desktop Overclocker's Lounge [laptop owners welcome, too]

Depends on how I have the chiller routed. When the radiators are bypassed (which produces the coldest water) I have one D5 pump running. It is sitting in-line, on top of the chiller. When I am using the radiators without the chiller, I have the 2 D5 pumps running. Using the radiators with the chiller running, all 3 D5 pumps are running. One pump is probably enough no matter what, but I like the redundancy. The Hydro Copper blocks do not seem restricted at all.

You will have fun setting it up. I am on a business trip this week, so only my phone and ZBook 15 are available until this weekend.

 

I run the GPU stock at all times except when benching. No reason to overclock it for anything else. I dial back the CPU to 50x with the adaptive voltage profile and balanced power profile when I am not benching. It runs nice and cool that way, except for when I forget to turn my fans back on, LoL. It gets warm pretty fast with the fans turned off.

 

Have a safe trip! Looking forward to catching up.

 

Don't I know that!! I did that a few times!

Stock is good.

 

Worlds fastest still with the new updated bios who cripple down performance to reduce Dell's RMA/Tech support costs Yeah, the Boss know how to "fix cheap flawed engineering".

https://twitter.com/AzorFrank/status/1135977863864303616

 

So not only is he excluding Clevo's from the world's fastest notebook title, he's even conceding that the GT76 will beat the A51m; the tag of "shipping today" is hilarious.

 

good, I'm worried I do not know if it will damage the imc of my i9 9900k CPU, can it be?

 

Sorry, I don't know if I saw this before. LTSC doesn't pack the full cancer payload, but it is still built on the same underlying feces code. In other words, it sucks a little bit less.

To put that in verbiage that is more succinct and palatable to the Kool-Aid drinkers, LTSC does not have the worthless Cortana, Store or UWP app rubbish.

 

Don't have as much time for this as I would like these days, but here another run:

https://www.3dmark.com/fs/19521970

Fastest valid Desktop 1080 Dual SLI run is hereby beaten by a Notebook.

 

Bloody hell, really, are you saying that you scored higher than any desktop GTX 1080 sli system out there, but you did that with a notebook? That can't be right can it? There's people out there with water cooled GTX 1080 sli for instance, not to mention LN2 and what not.

 

Counting only valid runs it pretty much looks like it:

https://www.3dmark.com/search#/?mod...=0&gpuName=NVIDIA GeForce GTX 1080&gpuCount=2

Desktop user may not have modded their firmware to optimize performance per clock...

 

with dual 1080s it could sure be possible. consider that guys playing around with LN2 / DICE / SS etc only go for the best, i.e. 1080 Ti SLI and not 1080s...

 

Most of them play with 1080 Ti cards

 

So which is the best then? Thanks for the reply brother. If you don't like the garbage, then why use Enterprise N?

 

Wow, it's true then, looking at that link from Prema! That's astounding, especially for a notebook - very well done! Prema, you talk about modding your firmware for performance per clock - what do you mean by that, are you talking the vBIOS or the BIOS of the motherboard? You modded it yourself, did you override 'protections' or 'restrictions', what was the general angle to improve the firmware? The only firmware modding that I'm aware of that might be related is "Mobile Pascal Tweaker" that unlocks the max power allowable for Pascal cards, so as to prevent power throttling.

 

Those TGP/TDP values have never changed since @svl7 located them first for the GTX6xxM generation.
Those tools just copied his work and automated the process so everyone could do it.
But just raising the TGP won't be enough to make a good vBIOS.
I've optimized the BIOS, ME, EC, EC2 & vBIOS according to personal systems ASIC quality...

https://twitter.com/PremaMod/status/874587630431858688

What I haven't optimized is this OS and it's eating up processing cycles...
Physics Score dropped liked 500 points by just switching the OS/drive...

 

Well, that's pretty cool, sounds like there's some secret sauce there! Does make you wonder why manufacturers can't just optimise it like this out of the box from the factory? I'm thinking that doing so might mean people pushing their systems too hard leading to RMA's for instance. To make full use of all your optimisations might mean ensuring that cooling & power capabilities are not exceeded, so perhaps special measures to be taken with regards to cooling and power delivery components?

 

I just installed a super cheap 1TB NVMe drive in my system, it seems pretty good and just as fast as the fastest (but that might depend on what kind of tests were run on it), I'm very happy with it and here are some more details:
http://forum.notebookreview.com/thr...ews-and-advice.429972/page-1133#post-10920840
I didn't want to include too much detail in this thread, because it's not directly related to overclocking, but I figured you enthusiasts might find it interesting.

 

Changed OS and did my first 1080 Port Royal run:

Spoiler

I liked the number so thought I would share it with the Team here.

Edit: Second run was only a 4 point difference:

Spoiler

​

This bench is pretty consistent when it comes to sucking for GTX.

Edit3:


https://www.3dmark.com/pr/101265

 

Windows 7.

Of the variations of Windows 10 feces OS, LTSC has the least amount of fecal matter. Enterprise N is missing Windows Media Player, and I use that as my media player. You can add it back in using Media Pack, but there no point in using Enterprise N if that is what you're going to end up doing. I don't care for Windows 10 multimedia applications or image viewers, and LTSC doesn't have any of that garbage.

 

I have just requested the hwbot admin team to consider a rule change that would allow us to bench mobile GPUs against their Desktop counterparts...worst case here comes nothing.

 

Would this be starting with 1080 and higher or all gpus?

 

I would suggest every card that has been given an official counterpart by NVIDIA.
So 980 and up.

Also not asking to remove the Mobile card sections. Just let us upload to Desktop section as well if we think our runs are up to the task.

 

Hummm, wonder if they would allow for multiple points when doing it this way.

 

Bumped my 2133Mhz ram further up to 2800MHz and with some changed timings.


Cinebench R15 score increased a bit (stock clocks). But not so much as I hoped. Still, 2110/2111cb is quite ok from an 4 years old notebook.


@Prema You have seen this colorful Desktop wallpaper before? My old eyes go totally wacko if I look long time on it, HaHa

Spoiler: World's Fastest and most upradeable laptop, according to Alienware's top duo Azor and Umar

How benchmark scores scale with different ram speed and timings. Not from 9900K but 8700K should still give an ok pointing stick how ram can affect scores Intel i7-8700K Coffee Lake Memory Benchmark Analysis Not put in stone but more like a guideline.

 

An ideal system would be to auto-add these GPUs to both sections and generate a score based on their position in each section. But surely no-one would alter the underlying code to do that unless they really care about the future of the bot.
No new Notebook scores for weeks even though we stepped back and I only added the two FS and TS runs.

UBI E3 stream is starting soon...WD3 here we come.

 

X299 Rig Update:

Got the two KPx Hydro Copper blocks in. They are currently sitting on my desk untouched while I wait for my VID connectors, D5 pump, nine Noctua 3000 120mm fans and an assortment of fittings to arrive. I haven't benched anything in almost two weeks and I have an itch I can't scratch.

 

I honestly don't think that Port Royal utilizes a GTX card properly or rather that its Ray Tracing performance is intentionally being slowed down on driver level. I mean look at the clocks you can run that bench at without dropping a sweat:

https://www.3dmark.com/pr/101962

That's like running GPU-Z render test. Same GPU can't run FS above 2113Mhz...

 

What about your 2080. Does this also do the same thing?

 

I forget where I either read this or heard it from someone. Maybe one of you guys can tell me if there is any truth to it. "The lower the last 4 numbers in the serial number on the CPU the closer it was to the center of the wafer thus a better overclocker". IE: 0062 is better than 2512.

 

I have yet to to run PR on the 2080 (currently don't have it in the system), but with it's 2100Mhz clock-block there will be no way to know unless dropping voltage...

 

That's a good question....

okay, thanks.

Talk about pulling some watts...

 

Ah man I'm with you brother... It has been over 2 weeks for me with all the chaos (good life chaos lol), but dang it has kept me from doing much... my KP blocks are still in the original shipping box lol.

Looking forward to it all...

@Johnksss it's nice being able to see it pull all that power eh? I'm actually having an electrician come out to redo some wiring so that I stop tripping the breaker lol.

@Prema Great stuff bro..killing it with mobile cards... that's awesome...

@Papusan nice timings.

@JoeT44 hope all is well brother. What's new for you with your benching rig?

@Mr. Fox will give you a ring perhaps tomorrow to catch up? Looking forward to it...

 

Nice! Some days I have to use multiple outlets.

 

Right?! I've got extension cords running throughout the house HAHA! I have to roll them up everytime we have guests, but once they leave right back it goes.

 

What is the normal amp or max wattage from outlets in let’s say the living room in most U.S houses? I expect 110v net.

Here home 230x16A. Older houses most of 230x10A

 

15A/20A

 

555W on FS, wow, that's insane!

Congrats on running 2200Mhz on water!

 

ah that explains it...with up to 67% moar powah available per outlet this might not be necessary in European overclocking households

 

Thanks!

Aio was pulling 500W

Not per outlet persay. They are sectioned off throughout the house. Like certain walls and or rooms share the same circuit.

 

Looks like ideal conditions for extreme cooling. Only 7% relative humidity and the dew point is the freezing point for water.

 

3DMark is turning into an advertisement for new hardware/features:

https://benchmarks.ul.com/news/new-...379.1489484916.1560393101-59954174.1560393101

Poor MSI...they shouldn't have waited so long.

 

I noticed my 1080s are woefully underutilised in the Quake 2 RTX demo. They barely pull 150W at "100%" core on both and give maybe 1fps improvement from 275W TGP and corresponding gpu/memory overclocks.

Could be something bottlenecking internally in the core, cache thrashing or something. But my GPUs seem to suck hard clock for clock regardless. You & Johnksss get like 10% or 15% more GPU score than me on FS/TS. Is your vbios really that good ?...

 

Quake2 RTX plays nice on Linux, too. Doesn't look quite as good as Windows, but somewhat expected. It still looks great, though.
Running it in window mode here to prove it's on Linux...


Some of the reflections are missing on Linux. For example, at the very start of the game, the character's reflection is not showing in the glass like it does on Windows.

 

Looks like more stupid nonsense coming our way courtesy of the Green Goblin. If this is for real, then to hell with NVIDIA. It will be really interesting it if matches or defeats a K|INGP|N 2080 Ti. That's sure to go over like a lead balloon with the folks that held out even more for that.

 

On topic of PCIe 4.0, it seems like an unnecessary new standard currently, Intel themselves have shown that it won't matter when it comes to gaming:
https://www.legitreviews.com/intel-shows-pci-express-4-0-means-nothing-for-gamers_212350

True, but I think anyone buying a current gen RTX GPU should have known already that they were being taken for a ride - price/performance ratio and disappointing generational increase in performance over Pascal combined with generally disappointing ray tracing performance - it was always gonna be RTX adopters being shafted by this one! Next gen RTX is gonna be the more sensible one to get (7nm)(not talking the upcoming "Super" GPUs) - I'm kinda locked into NVidia due to having a G-sync screen, I'm on the fence about RTX even for the next generation (7nm), not sure what I'm gonna do for my next GPU purchase.

 

But, what you are talking about is different in principle. It's one thing to be willing to adopt the new technology as the customers. It's a different dynamic when you take those customers that went all in for the top offering and pull a fast one on them. 2060, 2070 or 2080 "Super"? Sure, why not? Those customers intentionally bought something less because they were content with something less. 2080 Ti "Super" is a major violation to what should be viewed as their best customers and shows tremendous disrespect toward them. They know damned good and well those customers bought it, on purpose, because it was supposed to be the best thing available. Surprise, surprise, surprise...

 

What did MSI do? I missed that part.

Waiting to see if the voltage is going to be higher than 1.093V.


Also to add. Didn't they skip past PCIE 4.0 and right to 5.0 PCIE? At least that's what I heard.

 

I missed whatever happened with MSI as well. I was curious, but forgot to ask what @Prema meant.

I heard the same thing on the PCI-e. Don't know how much is real or rumor. Hopefully, the 2080 Ti / Ti Super bait-and-switch thing is just a rumor. What would be so odd about it is they don't have a compelling reason for it as far as I can tell. AMDs best still can't touch 2080 Ti from how it looks, so I don't know why they would feel compelled to crap on all the people that just bought RTX 2080 Ti GPUs like that. If AMD were putting the smack-down on 2080 Ti, then it would make sense and they would need to do something to maintain their decisive performance lead.