*Official Benchmark Thread* - Post it here or it didn't happen :D

[Mr. Fox]

11 hours ago, Papusan said:

When you are at it could you please check out current MC version added for all tested bios versions.

Apex Encore BIOS 0080 is 11F and so is the Apex 0071. What MC is the Encore 0071 you are using now? I switched back to 0080 on the Encore because I don't feel like taking time to tune away the memory errors for the 8600 profile. I may switch over to the BIOS chip with 0071 in a couple of weeks when I have more time to mess with it.

11 hours ago, Papusan said:

WTH is this? Whom want this in their pc builds? It looks like an mirror with all needed parts and screws taken from a big truck🤯 

https://www.techpowerup.com/317604/hyte-releases-its-first-all-in-one-liquid-cooler-thicc-q60

The CPU block mounted screen is idiotic, to put it bluntly. But, nothing is too tacky for gamers. Tacky and garish trash seems to be something they thrive on. But, screen aside, I like the fact that the waterblock is actually a waterblock, and the pump is mounted to the radiator. That is much better design for AIOs and a long overdue step in the right direction. The crappy pumps mounted on the CPU block is one of the design elements I dislike most about AIOs. It's antiquated and not a very smart design. The pumps are weak and noisy. Having wires running to the CPU block to run the pump and lighting crap on it also kind of sucks. If they made a version of this in black with no screen it might be something worth jumping on. 

[Mr. Fox]

I just snagged a good deal on an Aorus RTX 2080 TI Xtreme Waterforce GPU for Windows 7 benching. It looks almost new, includes original box, etc. It was the slave GPU in an SLI setup, so being watercooled and not used much (slave GPU) it should be a nice piece with a lot of life left in it. I did not realize these had 3 HDMI, 3 DP plus the USB Type-C. Crazy that you could run 5 displays on this.

 

 

 

 

[jaybee83]

damn, imagine this happening to you 😅

 

https://www.tomshardware.com/pc-components/gpus/buyer-purchases-rtx-4090-no-gpu-no-vram

 

[tps3443]

You guys are not going to believe this. Okay, so since getting my 14900KF I have been dialing memory just about daily. I have always had the issue of maintaining that true ram stability after rebooting or powering off my PC, I could get something crazy lkike 8533c36-49 100% stable for hours of stability testing, turn PC off and back on and fail in minutes. Last night I finally discovered what the problem was. I was able to repeat the problem right then and there. 

Windows power plan (Dynamic Boost with Disabled Core parking) this power plan will error out TM5 in 27 seconds flat. Whereas High performance mode or even balanced mode work perfectly fine for however long. 

This is not an included power plan for any Windows. So just a heads up for anyone running Specter OS. I have been chasing my tail and causing my own problems for the longest time. 

My mind is literally blown right now. 

I would avoid the power plan below with 14th gen. 

[Etern4l]

52 minutes ago, tps3443 said:

You guys are not going to believe this. Okay, so since getting my 14900KF I have been dialing memory just about daily. I have always had the issue of maintaining that true ram stability after rebooting or powering off my PC, I could get something crazy lkike 8533c36-49 100% stable for hours of stability testing, turn PC off and back on and fail in minutes. Last night I finally discovered what the problem was. I was able to repeat the problem right then and there. 

Windows power plan (Dynamic Boost with Disabled Core parking) this power plan will error out TM5 in 27 seconds flat. Whereas High performance mode or even balanced mode work perfectly fine for however long. 

This is not an included power plan for any Windows. So just a heads up for anyone running Specter OS. I have been chasing my tail and causing my own problems for the longest time. 

My mind is literally blown right now. 

I would avoid the power plan below with 14th gen. 

Thanks for sharing bro. This scared the hell out of me, I will just continue avoiding Windoze to be safe 😉 

[tps3443]

19 minutes ago, Etern4l said:

Thanks for sharing bro. This scared the hell out of me, I will just continue avoiding Windoze to be safe 😉 

I’m just glad I found the problem. I have been like thinking something is seriously wrong here. It’s insane when you spend hours or even whole day(s) to stabilize peak performance ram profile and timings. And you finally got it done, and out of no where it’s no longer stable. 
 

What OS do you use?

[Mr. Fox]

2 hours ago, tps3443 said:

I’m just glad I found the problem. I have been like thinking something is seriously wrong here. It’s insane when you spend hours or even whole day(s) to stabilize peak performance ram profile and timings. And you finally got it done, and out of no where it’s no longer stable. 
 

What OS do you use?

Just for giggles, try setting that power profile with the e-cores disabled and see if you regain stability even with that power profile enabled. I would be curious to know if the errors are only with e-cores enabled when using that power profile. 

 

Here is something else I discovered. If you are getting TM5 errors on a memory overclock that is normally stable, switch to a different CFG file, then launch it and switch back to the one you prefer to use. Something happens (not sure what) where it seems like a corruption of some sort causes the errors and loading the config file fresh makes the problem go away.  Where I have seen this happening MemTest won't have any errors and TM5 does and after a clean reload of the CFG file TM5 errors vanish. I can't explain why, only what fixes it. Maybe there are temp files that get corrupted and flipping the CFG profiles causes them to get deleted and created fresh.

 

Linux is really the only respectable operating system. It has limitations and isn't perfect, but it's not as screwed up as the other options. If Linux could run all of the software I want to use, I wouldn't even waste any drive space on a Windoze installation. 

[jaybee83]

4 hours ago, tps3443 said:

You guys are not going to believe this. Okay, so since getting my 14900KF I have been dialing memory just about daily. I have always had the issue of maintaining that true ram stability after rebooting or powering off my PC, I could get something crazy lkike 8533c36-49 100% stable for hours of stability testing, turn PC off and back on and fail in minutes. Last night I finally discovered what the problem was. I was able to repeat the problem right then and there. 

Windows power plan (Dynamic Boost with Disabled Core parking) this power plan will error out TM5 in 27 seconds flat. Whereas High performance mode or even balanced mode work perfectly fine for however long. 

This is not an included power plan for any Windows. So just a heads up for anyone running Specter OS. I have been chasing my tail and causing my own problems for the longest time. 

My mind is literally blown right now. 

I would avoid the power plan below with 14th gen. 

 

damn that is one of those crazy pc stories with a totally random solution at the end, love it 😄 

[electrosoft]

Jufes at CES talking to Jacob (formerly of EVGA). Good to see him in action!

 

 

[Mr. Fox]

11 minutes ago, electrosoft said:

Good to see him in action!

If you are referring to seeing Jacob in action, totally agree.

[Etern4l]

4 hours ago, tps3443 said:

What OS do you use?

Intel Clear Linux. It’s a less well known distro but beats Windows signiticantly in some native benchmarks, and certainly in practice.

[Papusan]

20 hours ago, Mr. Fox said:

Apex Encore BIOS 0080 is 11F and so is the Apex 0071. What MC is the Encore 0071 you are using now? I switched back to 0080 on the Encore because I don't feel like taking time to tune away the memory errors for the 8600 profile. I may switch over to the BIOS chip with 0071 in a couple of weeks when I have more time to mess with it

 

Same here. I flashed the 0071 only to see if I could get rid of the bug with CSM being randomly disabled with 0080. I haven't experienced this now with 0071... Maybe this come from my halfway faulty GTX 960. I swapped out that card after. I don't know. 

20 hours ago, Mr. Fox said:

The CPU block mounted screen is idiotic, to put it bluntly. But, nothing is too tacky for gamers. Tacky and garish trash seems to be something they thrive on.

 

Tacky and ugly is modern bro Fox. The more modern... The more ugly. And a lot people have awful taste. Hence they can continue make ugly things.

 

On 12/21/2023 at 7:50 AM, Mr. Fox said:

Add the Patriot Viper 8200 DDR5 kit to the do not buy list. This memory is trash. @tps3443found the same to be true in his testing. I was hoping to replace my generic A-die with a slightly better product, but it is more than slightly inferior. XMP1 and XMP2 both fail TM5 testing and increasing the voltage does not help. Tested with both the Z690 Dark and the Z790 Apex and the outcome is the same on both platforms.

 

It is ridiculous that the only DDR5 modules that have been good enough to keep are my generic green sticks and the G.SKILL 8000 kit I already own. Everything (literally) I have purchased hoping for something better has been utter trash.

 

We'll later see if this new mem kit follow same paths. Here's their new 9000 mem kits. I wonder what this will cost.

 

Why make it so ugly? Because this is the trend nowadays. Designers is veird people. They often have a taste tha's awful. But so is the buyers. 

 

 

Patriot gifts us Xtreme 5 RGB with red tops memory kit capable of 9000MHz, preview coming soon

 

Edit. Why run memory stability tests for days when a simple update can destroy your hard work after next re-start? @Mr. Fox

 

From Patriot's reps.... If your goal is +8400 stability in everything 24/7 then seek help. 

[tps3443]

4 hours ago, Mr. Fox said:

Just for giggles, try setting that power profile with the e-cores disabled and see if you regain stability even with that power profile enabled. I would be curious to know if the errors are only with e-cores enabled when using that power profile. 

 

Here is something else I discovered. If you are getting TM5 errors on a memory overclock that is normally stable, switch to a different CFG file, then launch it and switch back to the one you prefer to use. Something happens (not sure what) where it seems like a corruption of some sort causes the errors and loading the config file fresh makes the problem go away.  Where I have seen this happening MemTest won't have any errors and TM5 does and after a clean reload of the CFG file TM5 errors vanish. I can't explain why, only what fixes it. Maybe there are temp files that get corrupted and flipping the CFG profiles causes them to get deleted and created fresh.

 

Linux is really the only respectable operating system. It has limitations and isn't perfect, but it's not as screwed up as the other options. If Linux could run all of the software I want to use, I wouldn't even waste any drive space on a Windoze installation. 

Im just gonna stick with balanced or high performance power plan. Problem solved though.
 

She is bad too!

 

8533c36-49 with a 262k tREFI with 1.350v IMC. 🥹(I have a 240-250 watt cpu power limit in place for safety during this test on a stock CPU)

 

Now that I solved my problem, I can seriously start pushing it though.  Or leave it be lol.
 

1.332v load IMC voltage 1.350v in bios

1.300v IVR tx

 

[electrosoft]

3 hours ago, Mr. Fox said:

If you are referring to seeing Jacob in action, totally agree.

 

Definitely Jacob not Jufes. 🙂

 

[D2ultima]

22 hours ago, Papusan said:

 

So todays golden question for you will be... Will Intel do the same for 12th and 13th gen Cpu's as they did for the 14th gen? Aka only offer the APO feature for xx700 and xx900 SKUs ? This means you are out of luck if you still have the 13600K as I can see in your sig. This mean one step forward and two steps backwards for some of you. Hope that won't be the case. Sad if Intel trash down own i5 products.

See I agree with you here, but this CPU isn't soldered hooray!

 

That said, I think they'll eventually expand it to the i5 line, apparently it needs per-game testing and the reason both i7 and i9 get it at the same time is because their number of P-cores are the same. So perhaps EVENTUALLY the i5s will join. We will wait and see.

[Papusan]

59 minutes ago, D2ultima said:

See I agree with you here, but this CPU isn't soldered hooray!

 

That said, I think they'll eventually expand it to the i5 line, apparently it needs per-game testing and the reason both i7 and i9 get it at the same time is because their number of P-cores are the same. So perhaps EVENTUALLY the i5s will join. We will wait and see.

I hope you are right. But isn't the main goal with the APO optimization to make the baby cores work better in games? In short offer better thread scheduler to make better utilization of the low-power E-cores in Intel's hybrid mess. Yup, the i7 and i9 have the same amount P-cores but different amounts of baby cores. The only difference for the i5 is less of all cores. Not only the amount P-cores.

 

Intel shouldn't have such a fragmentation like this if they want to cater to the gamers. And many youngers buy the cheaper i5 models for playing their games. If they continue like this they just push more gamers on low budget, over to AMD. And If they continue like this they just shit in its own nest. They need to do a change and also include i5 chips from 12th gen and upwards.

[D2ultima]

56 minutes ago, Papusan said:

I hope you are right. But isn't the main goal with the APO optimization to make the baby cores work better in games? In short offer better thread scheduler to make better utilization of the low-power E-cores in Intel's hybrid mess. Yup, the i7 and i9 have the same amount P-cores but different amounts of baby cores. The only difference for the i5 is less of all cores. Not only the amount P-cores.

 

Intel shouldn't have such a fragmentation like this if they want to cater to the gamers. And many youngers buy the cheaper i5 models for playing their games. If they continue like this they just push more gamers on low budget, over to AMD. And If they continue like this they just shit in its own nest. They need to do a change and also include i5 chips from 12th gen and upwards.

What I have seen is that it actually specifically bans the e-cores in general from being seen by the games, except perhaps for some specific logic threads, which is why it's on a per-game basis. If that's the case then it shouldn't be too focused on e-cores but rather making sure the right threads for the game stick to the P-cores.

 

Some people were reporting checking the load balance and said the APO driver on/off toggle was an almost perfect process lasso effect to the P-cores. Take it with a grain of salt, because I can't personally check it, and I don't currently have the link to it (though I believe it may have been in here? But some tens of pages behind).

 

As for intel not having such fragmentation, I agree. However if their arc GPU software improvements are anything to go by (as @Mr. Fox will attest to) I would hope that this APO movement is under a similar branch where they'll try to work for it.

[Mr. Fox]

More often than not, Intel does the right thing. Working in an industry where the exact opposite is true, that says something. I think this is a Windows/software issue. Fixing it should be a burden for the Redmond Reprobates to carry. If their cancer OS wasn't such a piece of crap it would not need optimizations.

[Papusan]

33 minutes ago, Mr. Fox said:

More often than not, Intel does the right thing. Working in an industry where the exact opposite is true, that says something. I think this is a Windows/software issue. Fixing it should be a burden for the Redmond Reprobates to carry. If their cancer OS wasn't such a piece of crap it would not need optimizations.

Yup. And expect more changes to the power profiles to match their next gen Jokebooks coming this spring. One fits all bro Fox. And Microsoft deliver.... In loads. If it match their trashware.

[Mr. Fox]

8 hours ago, Papusan said:

Yup. And expect more changes to the power profiles to match their next gen Jokebooks coming this spring. One fits all bro Fox. And Microsoft deliver.... In loads. If it match their trashware.

It's not enough for them to castrate turdbooks with cancer firmware and EC herpes, they have to go the extra mile with collateral damage by removing their "frank and beans" with Windoze power profiles.

  

21 hours ago, Papusan said:

Edit. Why run memory stability tests for days when a simple update can destroy your hard work after next re-start? @Mr. Fox

From Patriot's reps.... If your goal is +8400 stability in everything 24/7 then seek help. 

His opening comments were very accurate...

"If you ever want to see a system get unstable fast, start installing RGB software."

---

 

 

 

[electrosoft]

47 minutes ago, Mr. Fox said:

His opening comments were very accurate...

"If you ever want to see a system get unstable fast, start RGB software."

---

 

 

I despise all this RGB harvestware. Razer sent me a pair of their Nommo V2 Pro speakers to review and while I think they're good, that Synapse is bloaty on an Asus level. I ended up just giving my system a clean install and opt to run them without it which means I can't control the RGB and OpenRGB doesn't support them yet so they are stuck in rainbow pattern.

 

Any RGB product should come by default set to OFF.

 

Of course how could they entice (compel?) you to install their software if you don't want it except by having everything default to rainbow cycle.....ugh.

 

$450 speakers and they can't even retain their settings?

 

 

 

[tps3443]

I think I’ll buy a 4080 Super for $999! If I can get my hands on one. I like some fast graphics memory. And I would not mind giving one a try to take advantage of the new DLSS features too. And for $1,000 bucks it is not that crazy. 

[electrosoft]

1 minute ago, tps3443 said:

I think I’ll buy a 4080 Super for $999! If I can get my hands on one. I like some fast graphics memory. And I would not mind giving one a try to take advantage of the new DLSS features too. And for $1,000 bucks it is not that crazy. 

 

4080 Super for $999 is a decent buy. 1-5% faster than the 4080 but $200 cheaper on top.

 

Outside of the VRAM, better than your 3090 in every aspect.

 

 

[tps3443]

Just now, electrosoft said:

 

4080 Super for $999 is a decent buy. 1-5% faster than the 4080 but $200 cheaper on top.

 

Outside of the VRAM, better than your 3090 in every aspect.

 

 

Time to retire the 3090KP! I work from home, and after 2.5 years averaging about 200 hours a month this thing has north of 6,000-7,000 hours run time on it easily 🤣

 

Im kinda amazed it hasn’t broken on me. Or maybe that’s not much usage at all. I’m not sure. 

[Etern4l]

17 minutes ago, tps3443 said:

Time to retire the 3090KP! I work from home, and after 2.5 years averaging about 200 hours a month this thing has north of 6,000-7,000 hours run time on it easily 🤣

 

Im kinda amazed it hasn’t broken on me. Or maybe that’s not much usage at all. I’m not sure. 

Depends on the load, but basically those cards are made to withstand a lot. One year of 24x7 is 8760h, those consumer cards can take many years of full load without breaking a sweat. Would not worry about it from the reliability perspective. The 4080S would indeed be a bit of an upgrade (ca. +20% since your 3090 performs like a 3090 Ti), unless something is memory-bound in which case there would be a 25% hit.