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

The voltage curve tool still works on mine with the Galax HOF XOC vBIOS. It does NOT with the K|INGP|N vBIOS. Perhaps you should try that one instead of the one you are using now. I get better stability and higher overclocks using that vBIOS than I do with the K|INGP|N vBIOS.

Nothing is reported accurately in software after the shunt mod. I think the most GPU-Z or HWiNFO64 have ever reported is maybe 350W. That's what the firmware is telling the software. The shunt mod disrupts "communication" between the hardware and firmware. But, you should be seeing a lot higher power draw from the wall. If your results are similar to mine the you should see around 750-800W getting pulled from the wall in a Fire Strike graphics test or Time Spy graphics test if the voltage is locked at the max.

Sure, no problem. That is the Galax HOF XOC vBIOS. (The ONLY version of the Galax vBIOS that is actually worth a damn. All the others totally suck.) If you flash that vBIOS you will have to clean install the GPU driver or it's going to be super buggy.

Thank you. I wondered if that might be the case.

 

Hummm, maybe that's it. Thought it was the XOC. Trying yours now.

With this bios I see a max of 365W/250W for second power.
On the other one I see 465W/435W for second power.....

 

Kingpin going for cheap now.
https://www.evga.com/products/feature.aspx

 

Well, I can't do any testing. I think my motherboard just kicked the bucket. It will not power up and has Q-code 00. BIOS flashback works, but it doesn't change anything. I see others with similar issues posting in the ASUS ROG forums and it is a defect/QC issue. What is weird is I went to reboot to switch from one version of Windows to another and when it power cycled is when it started.

 

RMA'd. Not going to do the ASUS crap QC again. Should have known better based on history.

EVGA Z490 Dark K|INGP|N Edition should be arriving on Monday.

 

Wow, sorry to hear that! That happen to me. I had to dry my board, remove all components and the reinstall them and no more 00. It happen the 3rd day I had the board.

Edit: I actually love this board. has everything I need for when i'm not over clocking.

 

Yeah, I really loved this board as well. Mine wasn't damp or anything, but the ambient temperature in my office is pretty cool and that could be causing it. (Yeah, I know that sounds really stupid.) But, it seems to not be uncommon. It may start working again tomorrow after it is warmer in my office. I saw several reports of it being thermally induced and has to do with wrong resistors or poor soldering. I found several accounts of this in the ASUS forums on different motherboards, and the video below. I still haven't forgiven ASUS for the problems I had before. I love their boards and love their firmware. Their customer service and QC sucks and I'm just not ready to deal with their incompetence again.

 

How many Asus XII Apex boards have failed? That's a Dominus board.

 

Yeah, that video is about Dominus issues, but same basic problem. I found posts about other models. This is my third ASUS board and all three have had problems. ASUS service sucks. They have no advanced RMA program and it takes forever to get things done. (Their crappy approach is mentioned in that video as well.) Three strikes with me, so to hell with them. They're out. Life is too short to deal with crappy companies like ASUS. EVGA wins. I'd have a replacement on its way to me today with EVGA.

 

De8auer with some early shunt mod work on 3080:

 

You did see my post two before yours, right?

 

Looking at the reviews on Guru3d for a number of AIB 3080 cards vs the Founders Edition - there's no performance difference when all are overclocked as hitting the same power limit by the looks of it. Hilbert didn't explicitly say that the power slider on each card only maxes to the same ultimate max wattage, but some of those AIB cards only have a 102% max power slider allowable position, whereas Founders Edition is 115%....yet comparing fps of those two overclocked cards showed exactly the same fps when overclocked - so that's why I've concluded that so far all AIB cards are gonna have the same allowable power limit (in terms of max watts) regardless of whether their power slider goes up to 115% or 102%, have you come to the same conclusion? Different AIB have different stock power limits and clocks, and the ones that have a higher stock power limit only have say a 102% power slider limit, whereas Founders is like 115%.....so all of this seems to point that all cards have the same max allowable wattage once the power slider is maxed out. If that is true, then the card I will buy is likely to be just based on price and noise....that's assuming I won't be doing a hard power mod on it. What do you reckon?

Gur3d reviews:
https://www.guru3d.com/articles-pages/geforce-rtx-3080-founder-review,1.html
https://www.guru3d.com/articles-pages/msi-geforce-rtx-3080-gaming-x-trio-review,1.html
https://www.guru3d.com/articles-pages/asus-geforce-rtx-3080-tuf-gaming-review,1.html
https://www.guru3d.com/articles-pages/palit-geforce-rtx-3080-gamingpro-review,1.html

EDIT: @ajc9988 , how much extra headroom in performance percentage did he get by doing a power mod?

 

I hear ya. Hopefully you'll get a good Dark board. And Asus Tech support is not the greatest, that's for sure.

Side note:
Just past Microcenter and the line was around the block! LOL

 

He got 59.07 frames after and 57.8 frames before the change, so around 2.2%. But that is not fair yet. He has not had a chance to make the card colder. He only had a chance to check shunt mod with auto fans.



He said he will have another video tomorrow to show better overclocking efforts. He just ran out of time to have this ready to drop this morning.

 

I hope so as well. I still cannot get it to boot. Hopefully, that is all it is. Extremely likely the CPU is the issue, but there is a remote chance it could be that. Especially unlikely considering the other examples across a number of models and generations of ASUS boards over the past couple of years. Some suggested it was a dead CPU is all of those posts, but all the examples I found posts about ended up being the mobo after swapping out RAM, CPU, and PSU with ASUS making them jump through hoops and representing that the problem was caused another component before accepting an RMA.

That said, knowing that you also saw it happen once with yours, I hope you do not see it again.

My gut was telling me to stay away from ASUS, but the EVGA Z490 Dark wasn't available until now. Luckily, I snagged a K|INGP|N version, which supposedly has some minor differences in components to Vince's specifications. Tracking info says it will be here tomorrow.

Besides this being my third ASUS mobo with issues, the last being the one with the fan controller that caught on fire and the run-around crap that ASUS made me put up with on that deal, taking nearly a month to send a replacement, etc. And, I look at how many Joe (Harlon) had to RMA for stupid things like dead RAM slots and similar things... I think he had at least five X299 mobo replacements from ASUS... in hindsight I would have just kept the Aorus Master.

 

Right, I see, thanks for the reply. Well I guess that means that power mod is not really applicable for air cooled cards, which is probably gonna blow the relevance of the power mod for me out the water?

 

Hummm, could be the cpu, but with out a supply chain of parts to test, it makes it totally worse.

Does it go straight to 00 as soon as you push the power button?

 

After a long night of waiting outside the Microcenter, I got my RTX 3080 EVGA Ultra card.

Stock out of box, no tuning, no messing with fans. Completely stock and beats my heavily OC, not stable 2080 Ti with ease. A quick game test showed nice performance gains at 4K. Very happy so far with this upgrade.

https://imgur.com/a/KKs2g61

 

Yes, instantly. It doesn't initially show other codes and then go to 00 after a second or two. As soon as I turn it on it shows 00.

The first time I did the BIOS flashback it ran through normal codes and tried to boot up, then turned off and showed 00 with the red LED. I think it is something on the mobo or something corrupted in the UEFI or the BIOS chip. Everything was working normal and I was in the middle of using it. It started at reboot, so I am guessing something with the UEFI. I had switched to a different profile maybe 2 or 3 minutes before that and loaded the wrong Windows installation. It was when I rebooted to switch to another version of Windows that it started acting up.

Interesting thread: https://rog.asus.com/forum/showthread.php?101472-How-to-fix-CPU-CODE-00/page2


I believe the ASUS and Gigabyte boards use a single larger capacity chip for dual BIOS and not separate SOIC-8 chips like EVGA does. I could be wrong about that, but it seems like that is how it is looking at information in their marketing material. Both ASUS and Gigabyte use a temporary contact switch to flip from BIOS 1 to BIOS 2, whereas EVGA uses the actual positional switch to change from BIOS 1, 2 or 3, more like the dual BIOS GPU.

 

I salute you sir!

Nice starting score indeed!

 

I think I'm gonna buy a 3080 too. I'd be curious to see your assessment as to how much it bounces off the Power Limit after you've upped the power limit to as high as it can go in software......and then overclocked without adding any additional voltage? Like if you run Graphics Test 1 of Firestrike Extreme or Ultra is always bouncing off the limit (historically one of the most power hungry "gaming-like" tests). And how about in GPU limited games/resolutions is it spending much time bouncing off the power limiter?

 

Let the games begin!

I think you are right. I just looked at my board can dont see side by side bios chips.

And that is an interesting video, something to look at forsure.

 

Lol, can't order the 3080 anywhere (websites unresponsive or no stock), looks like my trigger purchase will have to wait! £649.99 is the best price I can find....one of them happening to be that Asus TUF model which I assume is the one you linked in that video @ajc9988 . My rationale is that prices are gonna increase after the initial wave of stock on the market, and at least I don't think they'll get any cheaper.

What's our thinking on 10GB VRAM being enough for the next 4 years? I'm thinking it'll be alright for 1080p, that's without me analysing the situation beyond my prior knowledge that I accrued when I played more games at the start of this year. I'm quite excited about the new feature that will be coming that will allow the GPU to fetch & decompress data directly from the NVMe drive without any CPU influence/overhead.

 

I still remember not so long ago people were complaining about 8 GB ram being ridiculous and no game would ever use that amount.

 

Well, if what Moore's Law Is Dead heard is correct, Nvidia is doing a $50 voucher for cards sold in September. That means a $50 jump up in price after that.

On the other hand, look what happened with the 5700XT. It made the 2060 worthless and the 2070, which prompted the Super series to be launched. AMD has multiple big Navis to put out. I'm not saying it will win against Nvidia, I'm saying Nvidia very well could respond.

Also, if you watch certain reviews, like Hardware Unboxed and a video by AdoredTV (I can hear the eye rolls, but he actually thinks Nvidia will beat Navi, so hold your horses on him being an AMD shill for a moment), they showed some of those settings used to show large jumps over the 2080 were because they used the settings to need more memory than the 2080 had, but less than their 10GB limit. This was seen in Doom Eternal and a couple others Digital Foundry put out.

That doesn't mean it is a bad card, and it is enough for memory requirements for games now, but these cards are making 4K gaming more accessible, meaning more optimizations incoming and more focus to getting consumers using it. That could spell bad news with low memory count. But not in the next 2 years. In 4 years, maybe. That is also why they had the 24GB of memory to show off 8K@60 (with DLSS, so likely lower res upsample; when are they going to allow same res more detail like they announced with DLSS 1? Seriously, that is what I want, not the 1440 to be like 4K, but I digress).

So, if you cannot get one, do not be bummed about it. There is a good chance there are some more cards set to be released after Navi drops.

Also, rumor is the Navi cards have 128MB cache on the card, which is why they are only using a 256-bit bus. I want to see it in action to understand whether that was smart or dumb as dog poo.

AMD likely also has the direct from NVMe, as that was seen on PS5 and I think the Xbox Series X.

Either way, raises the question of the speed limit of the cards taking it from the NVMe. Also makes me wonder what 4x500GB PCIe 4.0 NVMe on a TR platform would do (imagine around 24GBps off those in raid feeding the GPU). Once they get that working well with the benchmarks, it could depend a bit on storage as well as GPU to get those top scores. But that is my hypothesis.

Edit: I also agree with some reviewers that 1080p may become obsolete between this and Navi.

 

What happen with the Nvidia launch guys? I stayed up all night long trying to buy a RTX3080, and the "add to cart" never became a option on the Nvidia website at all. And then around 9:30 AM I see the " Available 9/17 go away" then it shows " Out of stock"
I read somewhere that bots purchased all of them in milliseconds, because Nvidia doesn't have a CAPTCHA during checkout?

I've been waiting to grab one from Nvidia, and they literally never even became available or showed in stock at all.

 

We're seeing weird things with firmware on laptops lately as well. You can brick a laptop by setting voltage too low with the only possibility of recovering from that disaster being a mechanical SPI flash programmer. I would not have a surprised look on my face to learn that the same level of idiotic behavior is finding its way into the desktop space. Stupidity is like cancer... it has to be killed early, at first sight. Once it starts to spread and it is allowed to get out of control, fixing it becomes a challenge.

Yeah, you should have camped out at a Micro-Center, Fry's, Best Buy, or someplace like that to snag one like @Talon did. Getting one online on launch day is like playing a lottery.

 

I don't mind about 1080p becoming obsolete because I very much doubt I'll be getting a higher res monitor in the next 4yrs, so that will be extra headroom for the 10GB VRAM in my situation, because I'd only be concerning myself with whether or not 10GB VRAM would not be enough for 1080p over the next 4 yrs. 4K uses a lot more VRAM than 1080p, so depending on your 2-4yr VRAM prediction if that's based on 4K then I should be OK with 1080p for 4 yrs (postulation). Ah, your Doom Eternal example, is that at 4K too? If that was 1080p I'd be worried.

Navi could be an interesting one if I hold off buying a card that long (I don't really need one to be fair, so should be doable), when's AMD gonna drop a Navi that is competetive to RTX 3080?

 

The thing about doing the shunt mod without altering the firmware is that it is only a partial fix. It eliminates the crippling effect of gimped power limits, but it does not do anything for room-temperature "thermal throttling" GPUs are notorious for, or the castrated voltage limits. It takes a combination of power limit override and firmware exorcism to realize much benefit. Otherwise, you're just fixing one problem and ignoring the rest. That won't get you anywhere with overclocking fun. But, you would see a minor improvement. Keeping GPUs abnormally cold is the only way you'll get great overclocking results, thanks to the cancer firmware algorithms. Probably not going to get very far with the room-temperature thermal throttling cancer using air cooling.

 

Getting a 3080 for 1080p is such a waste, you’ll be CPU bottlenecked to no end. Now would be a perfect opportunity to upgrade the display to a high Hz 1440p.

 

On Navi, no clue. The release event is Oct. 28, so availability is after that.

 

Personally I would like to see 16GB on the next AMD cards. My 5700XT suits me well enough for now but thats the amount of vram I have been desiring for a while ever since using the 1070 and seeing the P5000 with its 16GB.

 

One rumor said $500 for one version of big Navi, and $550-600 for that same chip with 16GB. So you may get your wish. BUT, as I said, they are using regular GDDR6 (I think 16Gbps) AND only a 256-bit bus, instead relying on a massive cache system to make up for the lower bandwidth and slower ram chips.

Because of that, I'd DEFINITELY caution anyone thinking about AMD UNTIL reviews from 3d parties are out (which everyone should do for all hardware anyways).

Also, something I want to give Nvidia applause on, THANK YOU FOR GIVING REVIEWERS EXTRA TIME WITH THE HARDWARE AND THANK YOU FOR THE REVIEW THE DAY BEFORE RELEASE!!! That is awesome and transparent of them, because this morning, everyone knew what they were getting or standing in line for until the stores opened. This needs to happen by every tech company before every major release. Stand by your product. Be proud of what you make.

 

Gets me every time

 

I don't think that old video clip will ever be not funny.

100% agree with you on that. That should happen every time. Buying expensive parts on "faith" (for lack of a better term) that it's going to be good isn't ideal.

 

How Much CPU Does the GeForce RTX 3080 Need? tomshardware.com

We tested the GeForce RTX 3080 with six different CPUs, including a seven-year-old i7-4770K, to see how much your choice of CPU matters with Nvidia's latest enthusiast GPU.



If Roboo continue with his [email protected] he should surpass the Intel i3-10100 Processor



Btw. Nvidia completely killed Sli.

Nvidia has transferred all SLI implementation responsibilities to the game developer and game engine and won't release new SLI driver profiles for the RTX 20-series and older GPUs starting January 1.

The End of SLI As We Know It: Nvidia Reveals New Model

 

GN live OC of custom 3080s

 

Wow 4/8 cores for gaming is just about dead

 

Maybe for new stuff, but its a good thing we have 30-40 years worth of games that it does just fine for

 

Depends. 3080 is a 4K card. And with 6700/7700K running above 4.7 you should be good. But forget streaming/other tasks while gaming You can still enjoy your old rig with some OC love and 4K screen.

 

Ever since I tried 165Hz and G-Sync, I have been sold on 1440P. It looks like now in the later part of 2020, we finally have video cards than can handle our 2014-2015 era display technology haha. And 4K too!

 

Man, PC parts are so cool. I want a 4x8 set of these so bad!

This DDR4 4000 C15-16-16-36 kit available and it’s $430 dollars is just amazing to me! So expensive. I doubt I could get that speed with those timmings, but maybe pretty close to it.

I think the speed and latency is just awesome. But, they are a bit flashy?

 

Just a note (because I do not know your upgrade interval), Sapphire Rapids from Intel on Servers next year and Zen 4 at the end of next year or sometime in 2022 will both have DDR5. I do not know when Intel HEDT will adopt DDR5 (usually comes a year or two after the server chips), so you could have a couple years of use (or great to keep around, then buy used CPUs and MBs that use DDR4 and get hardware points, etc., if you are chasing points). I just wanted to give the heads up on timelines. PCIe 5.0, though, is not planned to be given to consumers for a bit, though, and PCIe 6.0 is already in development. PCIe 5.0 PHY can support Intel CXL which is platform cache coherence possible platform. CXL is HUGE. Hoping Intel isn't late with it because having the platform allows development for PCIe 5.0 devices. For PCIe 4.0, because Intel didn't have a platform, AMD was used for all development (source: ServeTheHome).

But, wanted to give a heads up on adoption of DDR5 by the industry in case that effects your decision.

 

Oh I’m just window shopping. But I do plan to probably sit on this platform for a little while.

I want a Xeon 3175X and a LGA3647 motherboard. But they are so ungodly expensive. Maybe whenever used pricing gets around $1,500 I’d buy one.

Such a silly platform I know, but for only coolness and nostalgic effect the CPU gives off.

 

While those shiny Royal sticks are not particularly attractive to look at (to me anyway) you'd likely be surprised at how they run. What they lack in curb appeal they will make up for in performance. I was able to run the Corsair Vengeance LPX 4000 sticks as low as CL14 at 4200MHz. Read speeds were faster at CL16 but copy, write and latency speeds were better with CL14 or 15. Timings that low at 4000 need more than 1.500V. That's really not a problem, but some people get weirded out at the idea. I used to run memory, in laptops even, at 1.650V for years with no issues doing so.

Mr. Fox's AIDA64 Memory Read Score 132324 MB/s 4200 MHz Corsair Vengeance LPX DDR4 SDRAM

 

I am blown away by this performance. Drastically faster than my 7980XE with my current DDR4 crap sticks. Absolutely phenomenal performance out of that memory set.

Did you sell this set?

 

Yes, Brother @Rage Set has them now.

 

Yeah, I think the power mod is only gonna be helpful with some water cooling so I'll be skipping that mod.

Yeah, that's true, 160 fps is ok for now! If I get back into competetive online multiplayer gameplay then 160fps avg is not ideal as I'd be wanting that as a minimum framerate, but it depends whether the game in question needs that kind of CPU power.....but for all other types of games I think I can get by with 160fps+, ha!

 

I am not that extremely familiar with all of the Samsung B die kits, and memory overclocking in general honestly. But I really need to upgrade to a better memory kit, and tune it properly. I was thinking there are numerous levels of B-die.

I had no idea that Corsair set was so good. What an amazing deal that I missed out on.