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

My babies are now reunited

 

omg...

 

Well, I decided to try something a little different... just for giggles.

Wanted to see how well I could tune a "daily driver" or "gamerboy" 5.0GHz (78%) overclock, mesh overclocked to 32x, with reasonably tight memory timings and the lowest possible CPU voltage using the latest BIOS (versus the K|INGP|N X299 XOC BIOS that I use for benching, which has no exploit mitigation whatsoever).

Here is what I am coming up with. I am actually surprised I could go this low on the voltage and still get such decent performance. Guess I better save this BIOS profile. Seems pretty good.

 

Pretty awesome Brother! So I sent the ram back and got a full refund.

 

Wow. They released the KP block and it's gone.....at 300 a pop.

Edit: Correction. It is not sold out. They placed the listing of the block with the auto notify and the block will be out sometime next week. Keep watch KPx owners!!!

 

If the Graphic card is 2k ... what is 300$

I admit I am crazy but never thought “we” are so many )

 

True but they added to the cost by including another OLED screen. As far as I can tell, the OLED screen is detachable, so they could have allowed you to use the one from the hybrid cooler. It is also appears that anyone can buy the KPx now. An associates code is no longer required and that's probably because they weren't selling as many as they thought they could.

 

Make 100$ on 10$ expense is legit

 

Here's a nice comparison of the amazing progress we're making with the help of the Redmond Retards. The Micro$lop Monkeys are the Masters of Rubbish.

 

That's a big difference in score for the multi thread test, a 41% increase for Windows 7! That seems more like a bug to me rather than a true performance indicator, I've never seen differences in performance that large, it's normally single digit performance differences at significantly less than 5% difference if I remember rightly when talking about comparing different tests on Windows 7 and Windows 10. Why is the "Multi Thread Ratio" different between the two tests, what does that variable mean?

 

I don't know. All I can do it look at the numbers like you, but it has always been that way for me since 1809, and RTX cards won't work with any Windows 10 OS prior to 1809. GeFarts drivers will only install with Windows 7 and 1809 and newer. (Won't install with Windows 8.X either. NVIDIA doesn't support those OSes for RTX.)

 

multi thread ratio gives you the ratio between single thread score and all thread score:

12183.7 / 593.2 = 20.54

so basically, 16 physical cores with HT at 32 threads gives you the equivalent multicore performance of 20.54 physical cores vs. single thread performance. so HT gives you an additional 28.4% performance over 16 cores without HT in this case. 20.54 / 16 = 1.28375

 

I am going to clean install Windows 10 1903 Pro and see if it fixes it. I doubt it, but will find out. My Windows 10 performance has been absolutely horrible since I had to upgrade to a newer version with the installation of an RTX card. wPrime 32m takes literally twice as long to complete than Windows 7 does. And, that is under both 1803 Enterprise LTSC 2019 and Enterprise 1903 (identical crappy performance).

 

@Mr. Fox I left you a message yesterday; I guess you were really busy at work. When I had 4x8 GSkill 4000 installed Windows 7 ran fan though not at the rated speeds and latency but when I did a clean install of windows 10 I kept getting one BSOD after another. Anyway I sent that ram back to Newegg and put my 3600 back in and found that 1 stick was bad. I know it was the ram and not the slot because I tested the bad stick in all 4 slots . So I'm running 3 sticks and have 2 more coming today but with only 3 ram slots populated I installed an earlier version of windows 10 and still kept getting one blue screen after another with different error messages each time. Windows 7 runs fine with 3x8 gb of 3600 so I'm really stumped. In the Apex manual it shows what slots to populate with 1, 2 and 4 sticks of ram but nothing about running just 3. So later today when all 4 of my 3600 is installed hopefully we'll see what happens. I just can't understand why windows 7 runs fine with 3 but 10 does not. BTW I ran that TestMem 5 for 30 minutes with no errors. Thanks Bro.

 

Something must have been messed up with the Windows installations. Seems normal now. Or, maybe the Digi-Nazis have done something to bugger up Enterprise. I installed Pro for testing. I will install Enterprise on another SSD and see if it runs like crap.

Yes, yesterday was a horrible day at work. (Nothing bad happened, it just sucked because there were too many people that needed assistance and I spent all day helping other people do their job and got nothing done that I needed to do.) By the time I was able to call you it was late on your side of the country.

I listened to your message and don't know what to tell you. Maybe there is something wrong with the ISO you burned to USB since Windows 7 is working fine. That, or the Redmond Retards are trying to download something that is goofed up during Windows 10 setup. Maybe try staying disconnected from the internet until setup is finished so it cannot download any updates.

It's 4:35 AM here. So, I am going to go to bed for a few hours. Never went to sleep last night.

 

That's more like it...a 2.5% performance win for Win 7 according to the two runs you showed there, no longer a 41% lead!

 

Some bench from the old lady Intel can go where some put in nasty things. Z170 still works ok with todays tech if you go the proper route.

https://hwbot.org/submission/416191...cy_core_i9_9900k_5103.72_mhz?recalculate=true


https://hwbot.org/submission/4161915_papusan_wprime___32m_core_i9_9900k_2sec_437ms


https://hwbot.org/submission/4161920_papusan_cinebench___r15_core_i9_9900k_2148_cb


https://hwbot.org/submission/4161923_papusan_cinebench___r20_core_i9_9900k_5045_marks


https://www.3dmark.com/3dm/36341689

 

Fantastic, Brother @Papusan. The immortal P870DM-G lives on yet another generation to continue its turdbook killing spree.

 

Wow, amazing job bro!

 

So after I installed the 4th stick of ram W10 went in without even a hiccup. That's weird that I had 0 problems on W7 and so many BSOD's trying to install 10 with 3 sticks of ram.

 

I am surprised as well. Sort of. Windows 10 is such a broken piece of garbage, I suppose that nothing should surprise us too much. But, I am glad you got that sorted out. Awesome.

 

Cool, that's a great upgrade and a shed load of power in a laptop - double the CPU power of my desktop in your R20 run, I got 2507, you got 5045! Am I right in thinking that 4.8Ghz is your daily driver OC, or you take it down to 4.7Ghz as shown in your R15 run, or maybe a bit lower even? Did you delid your 9900K yet, and were you using any special cooling methods for your benchmark runs there?

 

Well, another interesting discovery. The X299 Dark has two NICs and the one closest to the PS2 port (Intel i219V) is built into the chipset. The other one is an Intel i210AT and it leeches a PCIe lane. Seems like not a big deal, right? Well, it seems like using the i219V lets my CPU run better and faster. I get consistently lower CPU benchmark scores using the i210AT NIC. @Johnksss and @Rage Set you may want to check that out and see if the same applies in your situation. All I did was switch NICs and look how much my CPU performance improved... just at 5.0GHz, LOL. Going to have to try this on the chiller at 5.3GHz now.

 

Interesting find. The z390 FTW I'm running now also uses the i219V.

 

That is interesting. I haven't used that port yet, but will test it out after my Windows 7 install to see what happens.

Well damn! I went from 86.5 to [email protected]

 

Cool, what percentage increase in performance by changing to a different NIC?

 

I forgot to mention, although it should be obvious, that the offending NIC (i210AT) needs to be disabled in the BIOS. I will be interestted in hearing if it impairs performance on Z390 as well. Let us know.

Wow, that seems like an even larger improvement than what I saw.

For me it was about a 20% hit in performance in the CPU-Z benchmark, about 5-10% in other things. wPrime 32M improved from ~1.9s to about ~1.4s, about 800-1,000 3DMarks on the Physics test. But, looks like Brother @Johnksss saw an even larger improvement.

Now I should test to see if there is any change up or down in CPU performance associated with switching my M.2 SSD on port #2 from CPU to PCH.

 

So AMD will be showing us new (whether the product is of interest to you, doesn't matter) whereas both Intel and Nvidia will be showcasing rehashed improved versions of their products.

I am curious about the 9900KS, as I have a weak 9900K. The pre-binning should allow higher OC's. The Nvidia "Super" sounds dumb.

This is going to be an interesting week in tech.

 

Well the KS is an R0 (inbuilt security patches). Looks like a P0 with disabled mitigations autperforms an R0 clock for clock. But for the mainstream user who looks for security without loosing performance the extra Mhz should at least make up for the loss and then some...

 

4.7Ghz is more than enough for the DM model (oc is for benches). And yeah, I haven't delidded it yet. I always use mod Coolermaster U3 for my laptops (Mainly to avoid dust build up inside the chassis). You see... My wife's used nylon stockings can nicely live a second life

 

Hmm. Looks like TeamViewer has Linux support. Interesting.

 

Meh... not enough to matter. CPU performance is about the same either way. Seems like it's within a normal margin of error between runs. But, it was worth investigating just to be sure.

Now I suppose I should check CPU performance with both integrated NICs disabled in the BIOS to see if the i219 being disabled further improves CPU performance.

What I am thinking about here, and why I made the "interesting" comment, is I can use Linux as my garbage collection OS for day-to-day computing and maybe leave my Windows installations clean and optimized for benching. Since Linux has no registry and over-the-top dependency filth like Windoze, I can load it up with a crap ton of software garbage and have little or no effect on how Linux runs because you don't have all of the totally worthless performance-killing trash apps, services and tasks running at startup and corrupting the registry over time like Windows. I really don't need Windows for normal daily PC puttering around. Linux actually does a better job of that anyway. Windows is only necessary for benching and (to a lesser extent) gaming.

 

No way Jose, really! The NIC saps a minimum of 5-10% CPU performance with up to a 35% performance improvement in wPrime 32M!! That sounds wrong. I mean I'm not saying you're not seeing those results, but that implementation is fubar, how can a network card justify that much of a performance hit, I just can't see it. I mean how can that type of NIC connection be designed in such a way when the other NIC connection through the chipset frees up all that CPU power, and I guess with no downside to your network performance? Sounds like another bug to me, I can't imagine an NIC implementation being so fubar by design.

Sexy! Ha, but yeah, good implementation of that 9900K, that's quite an achievement.

 

I know, right? Pretty nutty deal. What got me to checking is that my Windows 10 clean install that "fixed" everything went away and performance dropped like a rock after a BIOS reset. I started trying to figure out what changed and spent many hours troubleshooting it. It adversely affected Windows 10 even more than Windows 7, so that is why we saw the huge delta between OSes before. And, that is where I landed. All three BIOS versions I have installed affect it the same, so if it is a bug, then I would say it is probably shoddy engineering of the NIC itself and not a firmware or driver bug.

 

Thankfully, it wasn't Windows 10 LTSC either. It's working correctly now as well with the cancer NIC disabled.

 

Wow such an overhead generated by the NIC...you should report that to the mobo ODM, so that they can update the NIC firmware if possible.

 

Yes, it's pretty crazy how bad it is. I think I will report it to them. Wanted to see if @Johnksss and @Rage Set could replicate the issue. Looks like John has, but I am not sure @Rage Set has had a chance to confirm it on his board yet. It would be nice to be able to point EVGA to this thread and say, confirmed by 3 X299 Dark owners. Not sure if they can fix it if the problem is the NIC itself. Intel may need to fix it. But, not a big deal if they don't. I am content to use the other one. I'm just glad I figured it out. This had been an ongoing issue that seemed random and puzzling. How the same overclock settings could perform 20% better or worse after switching to a different BIOS version or resetting NVRAM made no sense to me. I am always flipping amongst the 3 BIOS chips and applying different saved profiles for each BIOS version, and I am glad to know now why the performance has been inconsistent. I was never diligent about disabling that NIC and now that I know it was causing bad performance when enabled, I will be careful about disabling it.

 

I'm going to test this completely tomorrow. I do have two Fire Strike Ultra runs with the faulty NIC disabled.
https://www.3dmark.com/compare/fs/19435239/fs/19435422#

I'll run two tomorrow with the NIC enabled and alongside some CPU only tests like CB20.

 

Yep, I was about to suggest the same re contacting motherboard manufacturer, I see it as a bug, I can't imagine it's by design. They need to fix that!

 

Happy Memorial Day, folks. For those who have family members that have given their lives in defense of our great nation, thank you.
" Greater love hath no man than this, that a man lay down his life for his friends."

https://hwbot.org/submission/4163632_ | https://www.3dmark.com/fs/19438829

https://hwbot.org/submission/4163647_ | https://www.3dmark.com/vrpor/327887


More validation of the cancer NIC being disabled improves CPU responsiveness.
https://valid.x86.fr/es4e11

 

@Mr. Fox Great scores Brother.

What's amazing is that the one who said that laid down His life for His enemies also. Have a great day Bro.

 

I think the same, WW2 is where it's at for me in my mind in relation to that, my Grand fathers fought in that war - one didn't talk about it, and the other I still remember the stories, he was a bit of a Cad and a character, he had some near death experiences on more than one ocassion, and saved his life based on a gut feeling/intuition/premonition one time, he was an engineer in the army, died in 1995, many people had it harder than him in the army but I do appreciate their sacrifices, I imagine the world/UK would be a very different place without those people having fought for us. I'm not really a believer in dragging up the past nor holding grudges in any form, but it is good to honour the sacrifices of those when it made such a difference to our lives.

 

May those that stood up and fought against the evil in this world be remembered.
May we not forget, but forgive.
May the whole world be at peace, united in the brotherhood of man and the fatherhood of God.

 

Man, I am excited that this refined version is going to be released. Can hardly wait to play this version of the game that I have played more hours than any game in my life, LOL. Looks awesome. Coming June 6.

 

What do you use for games @Mr. Fox ? Some kind of joystick or controller? I've been thinking of getting a couple of games. Do you have any recommendations for beginners? Thanks

 

Any word on if it's gonna be playable on Pascal? I know Pascal isn't optimised for ray tracing, but as you know ray tracing is supported on Pascal, just it's slow. You'd think there might be enough overhead in this game to do it, given everything is literally p*ss easy to render in this game apart from the RTX reflections/shadows/lighting.

 

Oh, Heavens no, Brother. Keyboard and mouse FTW. Absolutely lethal against players using joysticks and controllers in a first-person shooter genre. There are games where a controller works better (Mortal Kombat) and a joystick (racing and flying games) works better, but I am primarily a FPS addict. I hardly ever play any other type of game.

Most of the basic controls follow a long-established standard for key mapping. W-S-A-D keys are typically directional movement keys. Holding down the shift while moving is usually for running (versus walking). Space bar is usually jump. C is crouch. E or F usually to "use" or grab something. V is usually for "victim" when you sneak up and stealth kill someone without firing a shot. G is usually for grenades and the top row of numbered keys usually associated with a weapon selection. Mouse movement is intuitive and the left and right buttons are the primary and alternate firing of your weapons.

I bet it will be playable. The Quake II RTX demo will be free, so you can test it with no cost involved.

https://www.nvidia.com/en-us/geforce/campaigns/quake-II-rtx/

@Robbo99999 - around 3:30 gets into the computational stuff, but the whole video is pretty interesting.

 

Bloody hell, raytracing related calculations account for about 90% of total frame time (while the traditional 'rendering' of the scene is less than 10% of it), and Pascal is over 6 times slower than Turing RTX at that 90% portion of the frame time. RTX is gonna have in the region of 6 times the performance in that game over Pascal I reckon. Assuming you might be able to run it at 100fps (optimistic?), then this means I'll be running at 17fps! Hmm. Well, it doesn't matter, but I'm just curious about these things.

That's quite cool you explaining the basics of fps gaming there - true! I've fps gamed since I was 16 (Wolfenstein!), and I'm in my early 40's now, so it's second nature to me. On the topic of fps gaming, I've spent a few hours in lobbies in BF1 with players hacking me left right & center (some days worse than others, weird) - I'm a bullet magnet for hackers because I'm MVP over half the time and it p*sses them off! This afternoon was an exercise in frustration or controlling thereof! I recently started playing at 180Hz rather than 144Hz, and it's a positive difference - crazy how much difference high refresh rate displays combined with high fps makes to your ability to play well. I turned down a few settings so I'm pretty much stable at 177fps in BF1 now with some drops of course - but still it's on maxed out settings at 1080p except no AA, Lighting Medium, Post Processing Medium (those are the ones to change for fps).

 

You should still try it anyhow. You might be surprised to find that it plays better than expected.

I have probably done less than 40 hours of online multiplayer in my lifetime. I don't really enjoy online gameplay much. What you describe contributes to me not caring for it, but I'd not care for it much even without any of that foolishness. The anti-social aspects of single-player gaming are what I enjoy most, and turning into a social event is the opposite of what I am looking for when gaming. That being said, I'd probably be fine with a private server and gaming only with a few select people that I know and like (with folks like you and others in this thread). That would be the internet equivalent of a LAN party. Haven't done a LAN party in years. Used to enjoy it with my sons when they were younger. But, I still enjoy the solitude of single-player the most.

 

Yeah, cool, I've never really done a LAN party either, at uni we did hook up 2 computers once though to play Quake against each other, but that's the extent of my LAN party experience! Well, for me online multiplayer gaming isn't really a social experience, I see it as a competetive and enjoyable/immersive experience - I'm not part of a clan and don't even have any added 'friends'(!), I just play with randoms as a lone wolf; but I do enjoy working with other players in the squad if they are playing the right game for their chosen class (e.g. medic/assault/support/scout) - but that's on a more intuitive level, no microphone orders or communications, I turn that all off, I have no idea what people are saying! (I do have the text chat on though and provide a few key strategy ideas if it's required on a certain map, or a thanks, or a heads up about an enemy being somewhere). I like the speed/quick reactions & quick decisions required and unpredictable competetiveness of it all, as well as trying to get better at it myself.

I'll try that RTX Quake when it comes out, see how GTX 1070 does....chug chug! :-D