BitMaster

Sure, all that applies. In the end I think cooling it with a waterblock might be the best solution of them all. Looking at that air-cooler monster, I doubt a card made by Nvidia WITH a waterblock would cost more, I think it's cheaper to produce a waterblock than such a massive aircooler.

Depending on your case, you could opt for an PCIe-v4 riser cable and mount the GPU vertically in the case and avoid the problem at all.

I think ED pretty much knows how much DCS needs Vulkan and Multithreading. There is no denial on that some kind of evolution needs to take place and ED has clearly stated that they follow the path of Vulkan and Multithreading to fix performance issues. How long it takes ? Nobody knows, not even the lead coder could tell you I am afraid. I am no coder myself but an IT guy and looking at the nature of DCS it is near to assume that it is not just complicated but yet predictable but complex, a totally different nature of a beast that every big code gets that has been carried along for many years, the 911 of flight sims, never changed really, only got better. You turn here and bang, you fix it and break two others, together with a transition from DX11 to an OS indep. Vulkan platform, great. I am happy they are at least working on it and not just plain let it run out as it is and say, enoug is enough. But It really matters how you look at it. If you are totally at it, invested a lot of time and money etc..I get it. Just stay tuned, I am sure ED delivers when it's ripe

Thing is, if you buy 2x 16GB now, DDR5-First_Releases, chances are high you won't find a matching pair in a relatively short time when they have moved forward with DDR5 production. 2nd, right now, with 670 and maybe also the B-chipset, at this moment, they do not run with 4 modules across the board. There are ZERO 4-module kits on sale, among other promised features still not available. If you buy 32GB now, you might end up buying a 2nd kit, 2x32GB whenever you realise 32GB dont cut it anymore. You wouldnt be the first one to learn it this hard way tbh.

Looking at those unboxing videos for 4090's...whatever case you get, GET A BIG ONE if this trend keeps going. I had to laugh...

Ah, ok new CPU, new terminology, new things to learn PBO2 is PBO on Ryzen5000 series, should have been clearer. There is no PBO1 and PBO2 switch. Your PBO is PBO..version2 so to speak. You need to know how to read that Auto/Manual/Disabled talk. Auto on my Gigabyte is PBO with CPU limits Enabled is the same but opens other menues where you can dial in higher values ( duration and power limits mostly ). Manual is everything at your disposal ( all above plus some more ) DIsabled..speaks for itself, no PBO at all. The easiest way to check what it actually does is to run HWinfo64 Monitor page while you stress test the CPU with any CPU hard hitting thing ( prime, linpack, etc. ). Your TDP is 105w, same as mine. With PBO on Auto and nothing else tuned hotter mine draws max 143-144watts, never more. Yours should behave similarily. If you dial in higher duration of peaks and allow higher amps for those peaks you can easily hit ~220watt with the same tests, and gain a lot of heat. The gain for gaming is next to nothing, it can easily force the CPU to throttle and YoYo up&down with the MHz more than if not overpushed. I think a constant stream of power is better than a maybe higher one in peaks but with fluctuations. The menues you say you dont see, become available mostly in OTHER tabs. The structure of the Bios is somewhat strange as foremost those CPU settings are spread across half a dozen of pages. My best advice to see how this works is a YT video with a Gigybyte board. I will link it. It's a very good method to squeeze more out of it while also undervolting. It''s very informative for Gigabyte users as he shows you the menue inside out in that matter. If you actually copy what he does is another thing. I did it all, it worked, lost the paper with the cores listed I took note of and then upgraded the bios....all gone. I redid it again etc.. and it's not worth it long story short vs. the power draw and heat. It's teaching a lot but I went with AUTO at the end of the day ( ~1 year testing what's best for me ). I tuned the RAM to 3600 14-14-14-34 at 1.45v. That is the only overclock I run. Tested it thoroughly and I can run VM's at 95-99% RAM usage for hours w/o any glitches or run linpack, aida, prime etc.. and no errors. If you oc the CPU, those will start to happen and verifying it's stable, and I need it stable, is a big thing. What you can do, is cool the CPU as much as possible. That will allow PBO to work at it's best That video is very informativ. I think I watched it at least 3 times, maybe 5tbh..LoL

Maybe I missed it and it's written somewhere, but do the 2 fans of the CPU cooler work LEFT-->RIGHT ( the other way round than normal ) or do they push air RIGHT-->LEFT as this would cause a less than optimal airflow with the rear fan? If your board has a Temp Probe connector, get a Thermistor cable and measure the temp at various places while you game. If the airflow is suboptimal it can lead to dead pockets where you have kind of stale air that actually only gets whirled around but not exchanged. The CPU cooler might be not enough for heavy all core workloads and stress testing, it lacks the mass and surface vs. a dual tower, but for gaming on a few cores with less than 100% load per core I dont think it matters. Before I would spend money on a DIY loop for the CPU, I would loop the GPU, makes 10x more sense these days. They burn 300w, the CPU is 50-80w, peanuts.

Wow, I am amazed how known and well used it already is. So, 2020 !.... and I thought I read enough IT news every day to not miss such goddies...lol...this wheel is spinning fast. I have to say, when it comes to Ad blocking, I dont need any additional extension as it blocks YT ads by default ( my #1 priority ) and the other sites look great too, filtered down to the last damn ad. The simplicity is great, I only needed LastPass and done. I have also only found 1 site it won't open, a client's IP-Phone WebGUI from Grandstream, those won't open with Brave. Yeah, installed on all machines, Win, Linux, Mac and going to bring it to clients as well.

A friend informed me about this new browser, based on Chromium as usual, that by default blocks all ads, incl. YT Ads, and much much more. I am using it for 2 weeks now and I must say, excellent. If you want a browser that doesnt need Adblock XYZ etc... try Brave Browser www.brave.com

Well, the whole story reads like what you say....and then "DON'T glue the Spreader back on !!!" It will need to be Direct-Die cooled I personally delidded, and glued back the IHS, on my 8700k and it still runs great every day as my HTPC with a Noctua fan meanwhile. With this, you really need to think twice. If you have ever worked on AMD Thunderbirds back in the day, oh oh, you know that those dies don't like any mishaps, not even the slightest. I smoked one with the first push of the button. I'll never forget that smell and grey smoke. Der 8auer's video is nice, seen it, but boy, there are many many tiny little parts everywhere. If one goes blingggg...ooops...you are screwed. It makes it interesting, rewarding if successful and thrilling all at once, just don't mess it up. Never, when you work with the CPU and cooler !

I wouldn't bother manually overclocking the CPU. Enable PBO2 and when it asks, CPU or Mobo limits, say CPU, that will limit the power draw to the TDP listed by AMD and not the higher values from Gigabyte ( duration of peak power and max Amp limit ). For the GPU-Aid. I used a Balsa stick from my Balsa stock ( yes, I used to build Balsa R/C planes, I guess I am old ) and cut it as long as needed. Any Pencil will help for now, cut it to length and see if it helps. I could lower temps by 10°C just by lifting it level with the finger. Instant drop of temp. Even if your temp doesnt suffer due to the hopefully better internal structure of the EVGA card it still helps the slot.

Sh!t happens ! call yourself lucky, you didn't drop the whole damn thing or fried the CPU, broke some LGA pins or spilled a Coke over the GPU, all would be way worse. When I did one of my rigs back in 2015 with the then new MoRa-3 ext. radiator, I accidently mis-connected the 4 180mm fans. When I turned it on 1st time, SMOKE...ehhhh dang. Turned out lucky, only 4 Fans, 50€, and a fan hub, 10€ an 5 days waiting for replacement. Shortly after I spilled hot coffee with milk and sugar all through the tower, top to bottom, through the GPU ( ugly mess ) while it was ON. Pulled the plug, took it ALL apart, bought special cleansing cans to spray-clean the PCBs...lots of work...but nothing broke, just a shock and lots of work. Ever since, my tower sits on top of my table, no more coffee from top to bottom HAHA. Call it a day, you sacrificed a door panel ! If the rest works as intended it's still a good day. Could have been way way worse.

Ohhh, Phanteks case....ehhh if that case has an LED controller, DON'T connect it to Gigabyte boards !!!! THAT was till today the only issue I had with my rig until I found out those 2 are incompatible ( newer Phanteks devices seem to have this fixed ). They don't show the Gigabyte Logo in that matter, compatible to "most" boards, Asus, MSI...but no GB. If it's connected, it wont boot. EDIT: yours is a NEW one, it is Gigabyte compatible and shows the Logo from GB ! anyway, better safe than sorry

Top Job

Clean build, nice to look at. You reverted the airflow back to front and top to front? As long as it stays cool I guess it should work but I would turn it around. I dont think the CPU rad will produce much hot air during gaming, rather use the natural way of hot air flowing up. Anyway, if it works I guess it's ok. If I was you, support the GPU. It really hangs downward. I had that on 2 of my cards and one of them it also heavily induced cooling issues as the card itself bent too muchand cold plate-PCB did bend away from each other. Put a stick under the point where it reads XCS at the far edge of the card. Watch temps under load before and after. I didnt believe it either until I tried it. It also supports the socket/mobo. I meanwhile run my card with a riser cable vertically mounted, just because of the weight and bending...and it looks great too. Is the RAM working OK now ? I wouldnt consider manual oc'ing the CPU, enable PBO with CPU default TDP. YOu can gain a few % with manual per core undervolting but it's a PITA when you find out every other Bios upgrade ruins your config..LoL I gave it up and my CPU stops at 144w. It can pull 220+w and do 5.05GHz single cores for a splitt second and maybe 200MHz more all core but the gain is minimal for gaming purpuses. Actually, that makes it easy to call it a day and forget overclocking the CPU manually. It also hardly ever goes into the ~70°C under stress.

How did you solve the rad problem ?

Well, that specific PSU is one made for that so it does come with the 16+4-Pin cable, ATX 3.0 specs and all that. But listen, even then and ONLY then you will encounter the melted connectors. Adapter cables top out at 450w and likely won't not melt away as the 600-660w one does. The big deal is, that it can destroy the conn at any device involved.... 1300€ GPU, 300€ PSU, neither one is cheap

The difference between 3080ti and 3090ti is down to ~100-200€ @ mindfactory for certain models. If I had to buy I would take the full thing. 12GB more for 150ish€, always ! https://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+RTX+fuer+Gaming/RTX+3090+Ti.html https://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+RTX+fuer+Gaming/RTX+3080+Ti.html

Let me bet 1 beer that this damn connector will see a Revision With such strict consumer protection laws in the US it's bound to cause legal issues sooner or later. I hope before the first home goes up in flames

I am no log wizard but it always says DX11 backend error... When did you last upgrade your Nvidia driver ? The only time when I encountered such things was when I tinkered with VMware or test-installs on physical machines with too little RAM, that often led to not shutting down properly. If it rather seldomly happens with small missions but happens often after larger missions it could be RAM, though 32GB should kind of be enough to not fail but who knows.

You say your SSD is near full ? Some SSD's firmware render the drive abnormally slow when near full. You might want to look into that as well.

With 4 sticks, you often have to raise Volts to like 1.375-1.4 Volts. As long as it is B-die Volts dont harm at those values but other vendor's chips may do if you stray north of 1.40v Try to find out what chips your sticks run on. I have B-die and they sweat at 1.50v ( set to 1.475v and the board adds the rest ). They are fine with that as long as you cool them aka keep an eye on temps. IIRC, Buildzoid said that the "other" vendors chip would die an instant death at anything beyond 1.4ish volts. Forgot the name of that specific chip that hates higher volts.

A hail to SSD & NVMe ! Coming from way back when CPU's had no Co-Processor and HDD had 20MB capacity, they, the HDD's, always were a very hard and costly to overcome bottleneck. For about a decade, mid '90s to mid '00 I switched to SCSI, Adaptec era, AIB Controllers and 4MB-Cache cards made my life easier but still, it was a bottleneck. Sometime around 2008 I built my last home-server with 16 drives, 14x250GB and 2x1TB, all on a 16 port Adaptec SAS BBU controller. Only to splitt I/O between OS, DATA and VM's. Gawd that thing was heavy, LOUD and overall not satisfying. I sold it and went Laptops for many years until I built my self another Do-it-All rig. That time around, SSD were already the # norm, WHAT a relief. That was around 2015. Now, I run ALL of my VM's off one single 1TB 980Pro until my 64GB are 98ish% utilized, that is usually either my UTM Firewall + all Linux + NAS or the UTM + 4-5 Win-OS There is no slow down whatsoever, I imagine if I had more cores and mostly more RAM I could fire up another 2-3 VM's and the NVMe would still just do it and not become a bottleneck. That sheer power is very welcome, first time we have headroom when it comes to drives and I/O speed. Unless you really need server grade throuput, it is almost impossible to bring a NVMe to it's knees on a desktop computer. They really strive when you press them hard. Small, silent and less heat too.

The DDR5 show is so bad that many shops and retailers bundle 32GB-DDR5 to a AM5 Board for free, just to push the whole thing forward. Most of the DDR5 promises made are still not here yet, neither could you do 128GB with confidence and sure no promised 256GB. And what did they say about Volts....I see the same volts applied as with DDR4. Maybe in the server space that comes to play already. 5800X3D, 64GB DDR4 , no stress...none of those new CPU's is actuallyfaster than that AMD chip. Many reviews skip it because it would show how badly they all perform vs. the 3D chip.

Also some boards for AM4 went through the roof in pricing. My board from last year sells for 600+€ now at mindfactory, thats 50% + to last year when I got it...LoL *edit: today that board is back to ~400ish€, what a freaking shock that was.