Mr. Fox

That's why I posted the Tom Logan video next to KitGuru because it helped bring some balance to the Kool-Aid drinking ceremony. When I first saw it and the opportunity presented to give it a go, I was initially skeptical because it is different. The overly positive KitGuru review allowed me to see it installed and now I am less skeptical than my first impressions. I will be doing a formal review on it within the next couple of weeks, so if there is anything about it that sucks it will be identified without any mincing of words. It is supposed to arrive tomorrow based on the tracking number I received. I'll put a link to the review in this thread once it is done. I will probably wait to begin testing it until I receive the 4090 late next week since this PSU is ATX 3.0 and has native 12VHPWR support. Maybe Brothers @Rage Setand @electrosoftcan tell us how this new driver is working out.

I think there is a good justification for this. If your space behind the motherboard tray and side panel is too narrow it might be an issue. But, think of cases like the Lian Li 011 Dynamic (including the mini and XL versions) and similar designs from other brands. The PSU could be the perfect candidate for some cases. The cables would point vertically and the PSU is installed vertically, with the fan pointing toward the side panel. A small case or a long PSU does not leave much space when the cables connect horizontally at the end and they easliy collide with other things at the end of the PSU. Seems like a really smart idea to me. It might not be in certain scenarios based on chassis design. I am looking forward to testing the arrangment and can certainly see the merit to it depending on the chassis. I am envious of your beautiful case (not the price of it) and you don't have to worry about some things like other people do because it is so large. I wish there were more tower cases as large as that one. It is unfortunate how small and cramped most of them are. I really hate cramped workspaces of any kind. Always better to have more room than what is necessary. You can ignore what extra you have that you do not need, but if you run out of space that just sucks.

I am also going to be receiving one of these new products. More to follow in the near future. I might actually be able to use the 3.5-inch drive cage/caddy in the basement of my Corsair case with this. (Can't with the 1600W PSU cables coming out of the PSU in the traditional fashion.)

My experience has been the opposite in some cases. I have needed to add voltage to avoid errors in memory stability testing at least half the time, even though most XMP profiles have sloppy loose timings and perform poorly. I think the reason they use such sloppy timings is because they are pursuing low voltage and care less about performance than keeping the voltage low. Also have to factor in variance in motherboard and CPU quality.

All laptops are nerfed, so don't feel too bad about it. Anything that is thin and light BGA rubbish is going to suffer from the bar being set very low and have limited potential. It's an inescapable reality. I know it is very frustrating to burn so many calories and see little or no change, and inconsistent results.

Good point, but it's also not smart to build or buy something like that in the first place. The problems with heat, power and build quality are a self-fulfilling prophecy.

Thank you. I still feel stupid for buying it last week, and almost cancelled the order several times since then. It has not shipped yet and I almost cancelled the order this evening. There are no Strix 4090 for sale anywhere from what I can see, and I don't want to pay an extra $400-$500 for one. I have only seen one available once and it was around $2499. I have a hunch that ASUS has temporarily stopped making them because sales are too slow and not many people are willing to pay stupid 4090 prices. I think all of the AIB partners are being conservative on production so they do not end up having excess stock they can't move like what happened with so many 30-series cards that are still available.

Nothing special to know. Other than it changing the hardware ID and name to "Galaxy" in GPU-Z it functions exactly the same. The only vBIOSes that don't play nice with other GPUs that I am aware of are the ASUS morphodite firmwares because ASUS HDMI/DP port mapping is proprietary and ends up leaving you with one or two display outputs that do not function. If you flash back the original firmware everything (including HWID and GPU name) will go back to original. PSA: I ordered a MSI RTX 4090 Suprim Liquid X and I am debating whether to put the 3090 Kingpin in my work computer and sell the 3060 Ti FTW3 or vice versa. I got lucky and snagged a "sold and shipped by Amazon" GPU for $1799. No third-party seller drama and no scalper fees. Probably going to sell the 3090 Kingpin so I can cover a larger portion of the cost of the upgrade. If anyone is interested, please let me know and I will put it up here in the marketplace or we can do it via PM. I always like to give first crack at my parts to people I know before I sell on eBay or Mercari. Having the mutual trust is a plus, and so is not rewarding eBay and Mercari for highway robbery on ludicrous seller fees and greedy Nazi "authorities" that believe it is OK to double-tax buyers for sales tax on used parts where sales tax was already paid by the seller on the new parts. It is still under warranty with EVGA. I just did a fresh application of thermal putty on the VRMs, new Snowman thermal pads and KPX a week ago. It has the XOC 1000W vBIOS provided to me via email by Vince (Kingpin) on the LN2 BIOS position, original box, etc. Excellent condition. I have the original hybrid cooler parts that will be included in case someone wants to use it on a system without a custom loop, but I currently have the Hydro Copper block on it and an MP5WORKS waterblock on the back plate. Before I sell the 3090 I want to make sure the 4090 is not a dud that I need to return to Amazon for refund or exchange. If it is poor to average silicon quality, I will let them have it back to pawn off on someone else. So, I am not in a huge hurry to sell the 3090 and won't even be ready to for probably two or three weeks if someone is interested but doesn't have cash tomorrow. I won't even have the 4090 until the end of next week, so no rush. I think Brother @Raidermanneeds this GPU so he can experience the goodness of ray tracing. 😜

I remember some of those things, but don't remember clearly what I did with some of them. Here is what I remember from the P870DM3 (never had a TM). keyboard lighting requires Clevo Control Center (CCC) cancer software SATA drives are a bit tricky to line up with ports and need to use something to reach under the drive and lift up If I remember correctly (big if) the BIOS fan option is to enable CCC to control the fans in Windows There were two 1440p screen options, one better than the other (sorry, don't recall details - I had the better one) Congratulations on purchasing the last good high performance laptop that was manufactured. The P870DM3 was the last laptop that I remember truly being fond of and didn't regret purchasing. I hope Clevo continues to offer service parts for it in case you ever need to replace something. Looks like this might have some useful fixes. Download AIDA64 Extreme (beta) Version: 6.85.6329 beta (Jan 31, 2023) Release notes: AVX-512 FMA units count information for AMD Zen 4 AVX-512 FMA units count information for Intel Cannon Lake, Ice Lake, Tiger Lake, Rocket Lake, Alder Lake, Raptor Lake improved DDR5 SPD bank switching identification of DDR5-9600, DDR5-9800, DDR5-10000, DDR5-10200, DDR5-10400, DDR5-10600, DDR5-10800 memory modules, DDR5-11000, DDR5-11200, DDR5-11400, DDR5-11600, DDR5-11800, DDR5-12000 memory modules PCI VPD detection / improved exception handling motherboard specific sensor info for Asus ROG Strix Z590-A Gaming WiFi II motherboard specific sensor info for Gigabyte A620 Series motherboard specific sensor info for HP 2B5E fixed: supported memory types list for AMD K19.1, K19.4, K19.6, K19.7 fixed: DIMM thermal sensor support for DDR5 memory modules fixed: MSI MEG Ai1000P, Ai1300P PSU sensor support fixed: sensor support for Aqua Computer LeakShield with FW 1019 fixed: motherboard specific sensor info for MSI MS-7D40 Nice to see you benching, Brother @Reciever. Since you said "not Kingpin" here is the Galax HOF 1000W 3090 vBIOS. (It also works fine on my 3090 Kingpin.) Rename a copy of the ROM to something short and sweet to make flashing easier. Also remember to backup your original vBIOS first. GALAX.RTX3090.24576.210125 (94.02.38.00.30).zip

Better off sticking with low cost options like Silicon Power, TeamGroup, Sabrent SSDs. The more popular Samsung gets the less quality and higher prices we see, with no meaningful benefits that I can identify.

I guess I never thought of it that way in terms of EVGA. Not only are they small, they are willing to break the rules and do things that Intel, NVIDIA and probably AMD do not approve of. They often make their competitors look bad and seem feckless, which probably does not win them any points among peers. My kind of company... not content with status quo and don't care about getting permission or blessings, or playing nice. I've heard rumors that their unwillingness to lock the IME and let customers downgrade or disable it in the BIOS is a no-no that Intel does not approve of. The other OEMs follow the rules on IME. I'm glad EVGA does not. One of the first things I do on my Dark mobos is disable IME. It's adds no value and is utterly worthless on consumer motherboards.

Yes, that is correct. Brother @Papusan is right. Even if the driver was not "approved" by UL (only meaning they actually tested the version) it would still be acceptable by HWBOT rules. There are no restrictions on drivers that are allowed with HWBOT. You can use any driver you wish as long as the results do not produce unrealistic benchmark scores that suggest that cheats were used. They can spot scores that are out of scope with an expected range of results because it tends to stick out like a sore thumb. That said, it is not a common scenario that any official driver version affects performance to a degree that the change in scores are freakishly out of line with the applied cooling and clock speeds. If anything, the scores will usually go down with a buggy driver, not up. General Rules | HWBOT The more common scenario is bugs and glitches in the benchmark software that produce abnormally high results, not the driver. If they spot glitched benchmark submissions where the score is freakishly high for the clock speeds shown in the required screenshots and validation links, they will disqualify and delete those submissions. In fact, there is some controversy of this nature relating to disabling memory ECC on 4090 GPUs. @Papusanand @johnksssand others have recently had their 4090 benchmark submissions arbitrarily deleted. In my opinion, this is being driven by golden boys having an issue with scores that are similar, higher or uncomfortably close to their 4090 submissions. They do not like it when peasants steal their thunder. For the first time, probably ever, I think the golden boys find themselves without any "secret weapons" and subject to the instability of the silicon lottery where GPUs are concerned and reacting to competition they are not used to having. With EVGA out of the GPU arms race, the level of mediocrity and loss of focus at the OEM level has never been greater. EVGA is not there to challenge NVIDIA's control freak nonsense and push the envelope anymore. I also believe that is why CENS was so eager to get his hands on the abnormally good CPU that @tps3443had. I am sure he was worried it would end up in the hands of a competitive peasant and that would be a threat. I really doubt he "needed" it for himself. He can get high quality silicon like that through special channels exclusively available to the golden boys that are chosen and predestined to stay on top of the leaderboards. They also get access to carefully selected and modified components and firmware that is not available to us pions.

To be clear, 18+4 is just ordinary gamerboy grade, not top shelf .

Brother @Papusanhas a sharp eye. The question mark at the end is the issue.

It can be either one (PNG or JPG) as long as the file size is not too large. I think it is 2MB or something like that. You will know because it won't allow you to upload a screenshot if the file is too large.

Yes, that is correct. You only need to copy and paste that URL into the space for a validation link and include that with the JPG of the screenshot.

Drivers have always been an area of opportunity for them. There have been periods of progress, but very inconsistent. Be sure that whatever driver you choose to use NVCleanStall to remove all of the worthless trash (GeFarts Experience, etc.) and do the MSI tweaks and driver signing. You only want the driver, PhysX and NVIDIA Control Panel. Everything else is a worthless performance-killer. Do not install the driver in the format that you download from NVIDIA. It will contain bloat and telemetry that hurt performance and offer nothing but stupid gimmicks as payment for allowing the data collection.

Probably not. Only EVGA and Galax have been able to skirt their stupid low voltage pansy-boy controls. That's why I recommended trying the Galax 666W vBIOS and Galax Xtreme Tuner software. There may be some low hanging fruit there. Only way to know for sure is try it. Probably not, but still worth trying to find out. Getting it up to even a wimpy 1.125-1.175V could help a lot. I have been taking my 3090 KPE up to like 1.450V without issue. The Suprim is built well enough to handle that I think. Some of those benchmarks I posted last night were also with a +1650 memory offset and +200 core offset. Thermals are the only thing supressing the core clock/boost. Without the chiller, the memory temps are pushing 80°C and with the chiller they were half that.

Use the script to permanently remove Windows Defender. Use ESO to kill all unnecessary processes (including NVIDIA Control Panel). Open the benchmark, then use Task Manager to kill explorer.exe and then click the button to start the benchmark. When the run is complete, hit Ctrl+Alt+Del and start Task Manager. Launch explorer.exe from Task Manager. Save your screenshot. That will likely give you the few extra points you need to breach 9K. Edit: I see you made it but didn't see your post. Try the above instructions and you might go even higher. Awesome stuff. Yeah, might try that Galax 666W vBIOS and see if it gooses the voltage a little more than 50mV. I doubt you are going to exhaust the power limits with such low voltage.

Cool. It sounds like it was a small addition of voltage, but actually enough to make an improvement. You might try the Galax, Zotac and ASUS GPU tuning software to see if it will give you the full 100mV. I would even try Precision X1 to see if it is compatible and what it can do for 4090. Never hurts to try. What are your memory temps looking like? I saw a video that showed the Suprim X Liquid memory temps were high because of the memory chips relying on tabs sticking out from the water block cold plate. They changed to ThermalRight high thermal conductivity thermal pads and shaved about 10°C off of the memory temps. They also moved from 1.0mm to 1.5mm pads (yes, thicker) for the memory ICs. Using thicker pads helped the memory but did not hurt core temps. This is also a good teardown video. The overall build of this GPU is impressive. Some very heavy duty stuff under that aluminum shroud.

In case anyone has experienced the same log-in failures I have at HWBOT Community, this might be useful. New login procedure for HWBOT Forum

Thank you. It's getting more difficult to enjoy the sport for reasons such as what you mentioned (quoted below). Tempting if you are a silly person. We often lean in the direction of silly when it comes to the strange compulsion to be on top when, at the end of the day, being a winner gets you nothing but a thinner wallet. Unless you are among the chosen elite that get cherry-picked products for free to facilitate marketing propaganda, then it's someone else's wallet.

The 3rd place out of 1426 submissions regardless of core count is certainly more meaningful than the hardware specific results. As people lose interest in benching--partly due to garbage tech, partly due to consumer ignorance, partly due to rigged scenarios with sponsors providing cherry-picked parts to chosen subjects and leaderboard politics--placement on the leaderboard is rapidly losing significance. It's a dying sport because of shifting priorities and probably the need for deliberate stealthy grooming of public opinion to distract consumers from the degradation of performance that goes with that territory. Thin, light, low-power have been made popular with a self-adoring population that excludes differing priorities. It is also worth mentioning that in most CPU and many memory-intensive workloads, Windows 7 steals the show. Windows 8.X, 10 and 11 are harmful to CPU performance. It is very difficult to match physics performance tests like those in 3DMark 11, Sky Diver, wPrime and Cinebench if you are competing against Windows 7 running a newer OS. In spite of the rhetoric about Windows 11 alleged thread scheduling superiority, it's not enough to close the gap. I do not think it is a coincidence that UL/3DMark is categorizing benchmarks like 3DMark 11 and Sky Diver as obsolete/no longer supported. It's because newer OSes can't compete when CPU performance is important to the outcome. It is not because they are no longer relevant from a graphics performance perspective because lots of modern games still use DX11, and lots of people still enjoy older games that use DX11 (and DX9, DX10 for that matter). So, it's ludicrous to take the position that it is not relevant. It's also an interesting "coincidence" that Cinebench R23 was deliberately/conveniently made to not run on Windows 7.

Exactly that, put to music. The lyrics describe this problem, and others.