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

Well I just finished running 5 rounds of Firestrike with different settings in the tessellation drop down, and I have yet to get an invalid score?? Since he was running the original AMD driver, maybe something has changed in the new one.

 

I think since this is not my best score with the R7 and I have the top spot already, it doesn't show.

EDIT: Beat my previous score: https://www.3dmark.com/fs/18845935

 

Ah, didn't know you could now alter tessellation. I knew you could do lod on firestrike, but not timespy.

 

Is that with your Primochill GPU water block? If so, what are your temps looking like, and how much higher can you go on the overclock with the Radeon 7 on water?

Please post some photos so we can see how it looks, too.

 

I just ordered the fittings yesterday... I didnt think it was gong to ship until the 25th, so I hadnt bothered yet. I do have the cooler, just no way of getting water to it yet. Should have it by mid next week. I will post some photos here in a few minutes

 

Here are some quick shots of it.

Sent from my SM-G960U using Tapatalk

 

Correct. With a lot less surface area of the direct die compared to the real estate of an IHS, the tolerance level becomes a lot more important on a direct die. This is very apparent as seen with laptops with its core differentials. This is why I lap the cold plate with laptop heat sinks in order to make sure it's perfect as possible due to the small die surface area, so you're theory is right on as validated from my flight hours of having worked with hundreds of bare die scenarios.

On the flip side if the contact point between the die and IHS is off, it could also have the same effect of a die and cold plate being off. The difference is that with a direct die contact it eliminates a variable of the possibility of the die -> contact being off, however, getting the contact points right on a die to cold plate is a lot more sensitive due to it being less forgivable.

Absolutely.

Amen.

Yes indeed. To add onto my reply above, the IHS being made of copper and with copper having a higher thermal retention rate among the common metals that we deal with in the realm of computers, if the thermal compound between the IHS and block can't dissipate the heat fast enough, there essentially is thermal back up occurring on the IHS. ie... the 'bandwidth' of heat is not able to transfer fast enough.

This is a very interesting topic that I've been looking into for a long time. When looking at XOC benchers, at those extreme temps with the right TIM there clearly is a point where the thermal dissipation capacity far exceeds the rate of the thermal transfer occurring during the benches. The amount of effectiveness starts to decrease right around going from DICE to the direction of the norm with chilled -> regular water...

I've analyzed a ton of bench results and when looking at regular water to DICE benchmark stats, the percentage of the thermal dissipation figures, percentage wise, sky rockets as expected, which has led me to the conclusion that everything (flipping point of the ratio of thermal dissipation) changes going from regular water to DICE, significantly.

Now that in relation to the TIM used comes to show that even with the very best regular TIM's (non LM), the w/mk is still not enough to effectively come close to a 1:1 thermal transfer rate. As you know LM drops off at around 10C with it being the limit, which has always made me wonder...gosh if Intel only soldered the 9900K correctly lol, we would be golden, but unfortunately their kindergarten solder job still forces guys like us to have to delid. (With the exception of XOC'ers using LN2..the solder is actually beneficial, although it's not the best solder job). It's good that we at least have an option to still take matters into our own hands so cheers to that eh?

Very nice brother.

Negative ghost rider. You're correct in that an invalid score is not eligible. I've had some of my best runs spit out an invalid result lol..total bummer so I never submitted it.

Nice. I'll be installing W7 onto my Z390 board this week as I think I've exhausted my regular water runs on W10 on this board, which did impress me but there's nothing like good ol' W7.

Well said, "I think that goes hand in hand with making sure that everything is as flat as possible, to me it seems like there's no room for errors when you go down that route."

With the bare die is a lot less forgiving, so being a little bit off translates to a lot.

Looking forward to it. Since the surface area of the die is small and with most water blocks being pretty flat, there often isn't a need to lap the block.

On the contrary with the IHS having more surface area and usually the bigger culprit over the block itself, lapping the IHS does bring greater benefits with lapping the block being an added bonus.

I wish more companies were smarter to support W7 ... this W10 rubbish is nonsense. Long live W7...

This is one thing that I wish 3DMark would change in categorizing better, because as it is right now it is complete rubbish.

People also need to understand that there exists different classes. What makes me chuckle is when I see people bragging or being arrogant about their benches and completely ignoring the existence of classes. (This is one thing that I really respect about you @Mr. Fox in that you are humble. Cheers to that.)

Ex. Most people just look at the total score and completely ignore the form factor of the system that produced the results. Take for example my desktop benches that I've been working on for the past few weeks. If the submitted benches were categorized correctly in a desktop class (on regular water) and not coupled with test bench results and XOC's results, a lot of my scores would be top in its class in respect to its set up and so would other bench(ers) submissions.

At least with hwbot you can do an advanced search, but most people don't take the time to do that. Even then a lot of those who claim to be on just regular H20 is questionable when you look at their graphics or cpu scores. There's no way someone on regular H20 is hitting DICE numbers, yet they submit it falsely in being in the H20 class...SMH...

Nice and congrats mate. I like the transparent intake and outtake section...neat.

 

Nice. That has some aesthetic similarities to my 2080 Ti Hydrocopper block.

 

Right before I had installed it into the desktop...



During gaming the GPU Maxes at 30C and when OC'ed 32C. It's an amazing block for sure...Heavy too eh? It's the big block of gpu blocks for sure...

 

New PR...

Knocking on 32K's door and that's my personal goal with this desktop.

Also, shot past my 40(500) graphics score goal.

9900K @ 5.4GHz (Physics PR: 28580) / 2080Ti
Fire Strike: 31806

 

@iunlock

You are getting getting closer.
28531K
https://www.3dmark.com/fs/18089754
29113K CPU Physics
https://www.3dmark.com/fs/17699792

 

Nice!

Once I hit 32K I'm pulling this card out and putting it on the test bench lol. It's a personal mile mark I so badly want to reach in this current form.

As for the 9900K this one is still non delidded...

What's driving me crazy right now is that I'm pulling pretty solid runs that should easily be outputting 32K, but it's being really wonky. I have a feeling it's an ecc issue as I was RAM tuning the last couple of days and perhaps something is off that I'm over looking...

Here's a run with a Graphics Score of 40795
https://www.3dmark.com/fs/18853194

Here's a run with a solid graphics and good enough physics that should be able to take it over the top (32K), but check out the combined...it's suffering I think due to an ecc issue... https://www.3dmark.com/fs/18853205

I'm just surprised that this non delidded 9900K is pulling 54x, but in these benches an error is negating its glory...

Hmmm.... I'll keep at it.

Update:
@Johnksss Just dialed in 31918 ... Almost ... Inches away...

 

Same with this chip. Non Delidded and sitting under ICDiamond.

The combined test is is weird one and you will need it in the 13K range to push past 32K I suspect.

You can basically tell if the combined test is going to be garbage right when it starts. The FPS needs to be a certain speed when it first starts to run. And generally ends in the 85+fps range on a good run, but that is 5.3 ghz. I'm pretty sure this chip will pull 54+ on desktop board, just not in a laptop. They seem to be locked around 5.2 ghz in here and 5.3 is pushing it.

 

Nice! Yea I've been binning my chips to get an idea to see where they stand before delidding... The combined and overall score with Fire Strike has been driving me up the walls lol. It's so random with a mind its own and just completely wonky. I don't get it... I think it's my RAM... so I'll be focusing that area now...

It's funny how when you do this long enough you start to notice things eh? I'm usually rooting when the benches start haha..

Now get this... I randomly run Time Spy before I was going to restart the computer to mess with the RAM and what ya know...

New PR lol..


Update:
Getting closer...

 

Finally! Lol ... 32K Club... All up hill from here...

This is a big personal accomplishment for me because I really wanted to achieve this in pure desktop form.

I always give the example of two cars at the 1/4 strip. Sure anyone can strip down a car to the bare bones (test bench) and run that 11 second down the strip. But to do that in a fully non hacked car is a whole different story and that's exactly what I was after... in not having to strip it down, take the seats out, headliner, side panels etc... :O


@Johnksss, it was indeed the RAM that was making it act all wonky. All good now as I think I've found a lot of the culprit in the ram settings.

 

About the Combined Score, I seem to be getting about a 2% higher combined score when I drop my core GPU overclock by one notch (13Mhz). This is over 3 seperate runs. Maybe your core overclock is borderline unstable and that's somehow affecting the Combined Score.

The reason I dropped it one notch is because I randomly found out that my long time core overclock has become unstable - at least in Firestrike Ultra Graphics Test 2. I've never run the Ultra version of that test before so don't know if it has always been like that, and I also changed to a new driver recently, although I personally have never seen in the past my overclocks become less stable with new drivers. Or I wonder if my over 2 yr old GPU is very slowly starting to degrade. Yep, so my long term overclock for gaming has decreased from a typical 2050Mhz seen while gaming down to 2037Mhz now. My GPU is not running hotter than before, so it's not that.

 

Maxon Offers Cinebench R20 Direct Download - Microsoft Store Not Required
http://forum.notebookreview.com/thr...nchmark-anandtech.827876/page-6#post-10887034

 

Glad to hear you got it all figured out!

You know. I use to be just like you in that respect,(Full desktop with no mods on air normal room temp and so on and so forth) till I found out I never came in first through tenth in world rankings.

But! I'm happy for you on the personal new record. Great job and way to keep it pushing!!

 

Yes indeed. The strange thing about Turning cards is that it has multiple personality traits lol. One minute it's fine then all of a sudden it throws a fit. I've spent a ton of time fine tuning the card and all is well now.

As for your gpu that definitely sounds like signs of degradation. It's really surprising how short the expected life span is for cards that are heavily OC'ed...which we all do. It's a pity that we're stuck with Micron chips, but hopefully that'll all change soon to the superior Samsung chips.

 

Thanks brother. It's always fun to reach goals and even more so when they are exceeded beyond expectation.

One thing about me is that I stay in my lane and just do what I do when it comes to how I go about testing / benching.

I put a lot of respect in honoring the different classes / categories in the benching realm, because like my car analogy at the quarter mile track, a stripped down (different form) drag car with no door panels, dash and only one bucket seat is not the same as a fully non modded regular car both running respectable times.

ie... it's more respectable when you see that fully clothed car running 11 seconds down the strip with a Slurpee in the cup holder, because it has one haha...

Therefore, when I see fellow benchers post up something I first honor the class and am just as excited about it as if it was a kingpin card, because in reality it has to be viewed that way since that is the reality.

Benching is awesome and I love the sport a lot. I've made several mentions throughout the years how benching is no different than regular gaming, but IMO much more rewarding and comprehensive than gaming. Amen @Mr. Fox ?

Uploading a bunch of stuff for points on hwbot isn't everything, rather I focus on being within the top 10% within the respected category/class. That's good enough for me, but sometimes it's nice to see my projects top its class, which is pretty cool, but I make sure to stay humble about it and just stay in my lane.

This gaming desktop build is just that and fortunately so far in the regular H20 9900K/2080Ti category it's sitting at top in some of the 3DMark benches/Cinebench, as a true to form desktop and not an open face sandwich (test bench). I love open face sandwiches too lol.. I have that as well for its intended purpose and I'll be diving into that next... errr after diving into the Area 51M that just showed up... lots of projects!

When it comes to my builds I focus first on practicality, because at the end of the day it's for gaming and work tasks. At the same time it's fun to maximize its potential in its current form and if there's a strong performer, it's even more fun to pull that card/cpu out and to throw it on the test bench.

Even when I bin my chips for myself and others, one of the things that I do is to bin them inside of a closed case (also) to mimic a real life scenario with ambient temps. To me this is very important, due to how picky chips can be at certain overclocked speeds, especially at common ambient temps...

All in all I'm grateful to share the experience with other fellow enthusiasts and to help others get on their feet to enter this awesome realm of benching...

.

 

Yes! Amen to that!

 

So tonight I'm spending time with the Area 51M... just barely unboxed it...

Here's the graphics score (fire strike) so far on this stock system... pretty impressive..



This is about to get realll interesting.

 

Please take pictures once the fireworks / smoke signals start

and DUDE, how much hardware are you currently playing around with at the same time???

 

Yea haha... It's expected to be in the red for sure and it is at 50x on all 8 cores... Stock it isn't too bad for what it's packing... Even some lesser H series CPU systems hit 90C+ out of the box when running a single Fire Strike run... but with this on stock clocks only one care hit 90C and every other core was around mid 80C's which is pretty good. The GPU is pretty respectable out of the box as well...in the 70C's ...

I'll be repasting this thing for sure...

As for the hardware, I've been really busy with the desktop build, while setting up my test bench with the z390 board...swapped out the x299. I'm always working with laptops and custom builds for folks so it's pretty much every day for me... great fun...

 

If it is indeed degredation of my GTX 1070 then I am still quite surprised to see it happen. In general I've been running it without added voltage - ie 1.062V max with 1.05V stabalised during gaming. The GPU is in the 60-68 degC range when gaming, which is quite low too, so that should help. If it is degredation I wonder how much longer the card will last, it's only degraded by 13Mhz over 2.5 yrs, so I'm thinking it will last pretty easy until my next GPU upgrade next year sometime. The only GPU that ever died on me was my overclocked & overvolted GTX 670MX in my old Alienware laptop in my sig, that was overclocked by 87% for nearly 4 yrs though, haha, now that was worth overclocking!

 

Do you sell builds then as a side profession/hobby? Certainly one way to get very proficient at working with PC hardware, you be building them with your eyes closed - Full Metal Jacket gun stripping style! (or maybe he didn't strip it eyes closed, he did eventually get to know his rifle well though, too well!)

 

It should last you until your next upgrade just fine. GPU's and even more so CPU's are extremely resilient. I can count on one hand the amount of chips in general that kicked the bucket over several years of data... if anything I've had RAM go belly up quite a few times, but these were some cheap ram and not Samsung or SKHynix or anything.

I take it that you have Samsung chips on your 1070? It's a fine card... in fact one of my favorite cards, especially in a laptop.

Haha yea I can pretty much disassemble AW laptops with my eye's closed now. It's a profession and a hobby. Repetition is the mother of all learning eh...

 

I got Micron chips on my GTX 1070, but I've flashed a vBIOS that has the "Micron fix", which allows for greater overclocking of Micron VRAM - previously it used to checkerbox crash with relatively modest VRAM overclocks. I got +180Mhz extra on the VRAM (MSI Afterburner slider) by applying that updated vBIOS - that was ages ago I did that. (So +400Mhz on the old vBIOS, +580Mhz on the new vBIOS).

 

Got my compression fittings today for the Radeon VII, the tubing, and the swivel connectors will be here tomorrow. Did a Superposition run, and logged the gpu temp, and the hot spot temps, so I will be able to better compare the difference. Should see quite a bit of a change, and hopefully a 10-15% increase in overclock. Hot spot temps were a blistering 113c

 

So the Omega board was DOA. Returned it today; already got a full refund. I'm back with my Rampage VI Apex which I really like. You know the old saying "don't fix what ain't broke" And the der8auer direct die sucked too. No better temps than before so I put the heat spreader back on.

 

On another note, @Mr. Fox I apologize for not calling you back. I fell asleep in my recliner at 5pm and woke up at 3am. That's what happens to old farts like me.

 

That sucks. I was hoping to see what the board was capable of.

 

So I spent all night last night trying to figure out the dreadful 0xc000007b error when trying to open throttle stop... Gee thanks windows *face palm*

I've tried almost everything... At first it was throwing a fit about two dll files... So I replaced those and then the error above showed up...

All that time I could have spent benching...

On the bright side, I'm close to hitting 30K (graphics) with this 2080 on the area 51M...next I'll just need to dial in the cpu to break 26k+ overall in Firestrike...

 

Haha, usually I am taking a leak between midnight and 3 am, so you cant be that old

 

You must have just been tired. I haven't slept for 10 hours in 30 years that I can recall, except for maybe the rare example of having a virus, flu or something like that (which is also very rare for me). Even 6 hours is abnormal for me.

I usually go to bed around midnight, wake up at 4:30 and then I head to the gym for my daily workout. I have been trying to force myself to go to bed by 10:00 or 10:30 PM because I know it is not healthy to live on only 3 to 5 hours of sleep per day. If I sleep 7 hours I usually end up being angry with myself for wasting so much time sleeping.

 

You mean break 30K total score in FS ? Good luck.

 

Ya, that might be a tough one to do. Rage Set did 27K with the Radeon VII and 7980XE, 30K graphics, 37K Physics, and nearly 13K combined.

 

¨Hence I said... Good luck

 

I mean graphics score ...30K isn't impossible... Overall score wise 26425 is the goal to top.

Ah forgive me if that was confusing I was extremely exhausted... but I had meant 30K graphics score, not overall.

Yea haha.. 28194 is the highest on Air, but that's on a regular mobo on a test bench so yea lol... I would have assumed that you knew that I had meant graphics score for 30K, even with my typo ;P

 

I kind of figured thats what you meant

 

Good day fellas, anyone at pax east?

Also, prepping for some benching today and installing windows 7 onto my z390

 

Several of my employees are going on Sunday. I'm working. Yay!

 

Started installing today. If anyone knows any of the 4 different Chinese languages that my instructions are printed in, please translate them for me.


Sent from my SM-G960U using Tapatalk

 

Long live Windows 7!!!

 

Exciting! Congrats...

Yes indeed! LONG LIVE Windows 7... Cheers to that.

 

Guys check this out ... just got in the new TX480 Rads from XSPC. It's the thinnest rad on the market and it performs on par with the thicker EX series supposedly...



Here is my USB Flash drive next to it for reference... crazy isn't it?



It's not has heavy / beefy as my Nemesis Rad, but we'll see how it performs...



Here it is next to one of my Nemesis GTR that's 55mm thick...

 

Yeah, massive difference in thickness! It will be interesting to hear any performance comparisons between the two if you're doing like for like testing.

 

Tried seeing Apacer, ADATA XPG,Crucial Ballistix 3600MHz RAM, Hyper X kingston, Corsair Vengeance 3000MHz+ all switched to samsung B Die and should perform similarly.

 

You do know that is only because I stopped there right? That score can go higher at almost any time.

And you still have quite a few people in front of you... Depending on how you look at ranking.
I personally look at the over all score because that decides the real winners.

 

Guys... here we go... another round of success and another milestone or achievement!

With my hard locked down Intel CPU 6700HQ... I managed to cross the hard limit barrier of 102.7MHz BCLK to 107.9Mhz BCLK!! thanks to the newer uCode or Microcode that I have used it & posted by me here... http://forum.notebookreview.com/thr...fix-and-meltdown.806451/page-17#post-10888780

I can go further with my so called ( locked down CPU) 6700HQ the official spec. on intel website is here... https://ark.intel.com/content/www/u...6700hq-processor-6m-cache-up-to-3-50-ghz.html

Here's some images...




Now I need to re evaluate & test my system or laptop including my RAM OC since BCLK will OC the RAM too further more and it is kinda unstable and throwing errors


Here we go again... another adventure to the OC world


I think if Intel saw this... I will be on jail