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

[electrosoft]

15 hours ago, Mr. Fox said:

 

Vulkan - GPU 645W power draw core clock 3135 @ 1.100V and +1500 memory offset / no chiller @ 82°F ambient

DX12 - 602W power draw core clock 3135 @ 1.100V and +1500 memory offset / no chiller @ 82°F ambient

 

Vulkan definitely makes a difference and blocking it always helps (look at those flat lines! lol). Let's see how much more the chiller helps on top of the custom water to really push it.

 

27 minutes ago, tps3443 said:

Pretty nice run @Mr. Fox I get exactly 8,020 with my 4080 Super on Vulkan API. 

 

How did DX12 hold up in comparison?

 

Has anyone tested Steel Nomad with scaling CPU clocks to see how much CPU contributes since it is the "successor" to Timespy where we at least got individual scores and a combined score?

 

[Papusan]

23 hours ago, Mr. Fox said:

Maybe God is trying to send a message to me that I need to stop wasting my money on computers.

 

Hmmm. Soon, maybe US citizens will be blessed with even more expensive hardware. 25% tax will soften  the need for buying new hardware. With 25% tax tariff you all will get Norwegian prices. Not cool.

 

PC gamers face higher hardware costs as tariffs on Chinese goods set to return

Graphics cards, motherboards, and desktop PC case prices could increase by 25%

[Raiderman]

7 hours ago, electrosoft said:

 

Vulkan definitely makes a difference and blocking it always helps (look at those flat lines! lol). Let's see how much more the chiller helps on top of the custom water to really push it.

 

 

How did DX12 hold up in comparison?

 

Has anyone tested Steel Nomad with scaling CPU clocks to see how much CPU contributes since it is the "successor" to Timespy where we at least got individual scores and a combined score?

 

Vulkan doesn't run as well as dx12 on Amd. At least that particular bench doesn't.

 

7 hours ago, electrosoft said:

 

Vulkan definitely makes a difference and blocking it always helps (look at those flat lines! lol). Let's see how much more the chiller helps on top of the custom water to really push it.

 

 

How did DX12 hold up in comparison?

 

Has anyone tested Steel Nomad with scaling CPU clocks to see how much CPU contributes since it is the "successor" to Timespy where we at least got individual scores and a combined score?

 

I'm curious how your system would do with that bench?

[ryan]

 

2nd with notebook 3060

[tps3443]

I had to share this! I have been trying to run such settings for some bit now.

DDR5 8600 with only 1.400 vdd/vddq set in bios!! (NOT 1.450vdd, NOT 1.435VDD but only 1.400V! 🥲 (All done on my G.Skill 8200 sticks) Its crazy how a really good IMC directly determines how good your ram is and how low the ram voltage can be set.  
 

This is such a nice little chip. I love it. Glad to be back on solder. 😎

[Mr. Fox]

1 hour ago, tps3443 said:

I had to share this! I have been trying to run such settings for some bit now.

DDR5 8600 with only 1.400 vdd/vddq set in bios!! (NOT 1.450vdd, NOT 1.435VDD but only 1.400V! 🥲 (All done on my G.Skill 8200 sticks) Its crazy how a really good IMC directly determines how good your ram is and how low the ram voltage can be set. 

Very nice. I haven't gone back to 8600. I used to run it all the time. I may try it again. I stopped only because 8400 with tighter timings produced the same performance and benchmark scores, but I do miss seeing 8600 just because many people can't do it and it is somewhat of a bragging right for that reason. It just feels good to run a system 24/7 beyond what most are capable of and I do miss that.

1 hour ago, tps3443 said:

This is such a nice little chip. I love it. Glad to be back on solder. 😎

That is a very nice CPU. But, why are you glad to be back on solder? I got my 14900KF from cletus-cassidy today (which he got from MarkDeMark) and it has not been delided. The IHS has been lapped, but it still runs hotter at 5.7 GHz with lower voltage than my bare die CPU does at 5.9 GHz with higher voltage. It is a great CPU and I am happy with it. But, I can hardly wait to get it on bare die. It's just way too hot with the factory solder unless I run it on chilled water. 

 

I think I am going to put one of my systems back on the open bench again. I'm getting tire of the vertical motherboard arrangement, so I will probably sell the be quiet! Dark Base Pro 901 and just live with the dust. I did get an air purifier about a week ago and it has helped a great deal with the dust in my office. I do love the Dark Base Pro 901, but the horizontal orientation of motherboards is just way superior compared to a vertical orientation. Every time I see the Praxis Wetbench in my closet I wish I was still using it. I'm so glad I did not sell it like I was thinking I might.

[Raiderman]

7608 on steel Nomad. Won't go much higher without chilled water. There is a bot submission of 8442 on a chilled 7900xtx.

[tps3443]

7 hours ago, Mr. Fox said:

Very nice. I haven't gone back to 8600. I used to run it all the time. I may try it again. I stopped only because 8400 with tighter timings produced the same performance and benchmark scores, but I do miss seeing 8600 just because many people can't do it and it is somewhat of a bragging right for that reason. It just feels good to run a system 24/7 beyond what most are capable of and I do miss that.

That is a very nice CPU. But, why are you glad to be back on solder? I got my 14900KF from cletus-cassidy today (which he got from MarkDeMark) and it has not been delided. The IHS has been lapped, but it still runs hotter at 5.7 GHz with lower voltage than my bare die CPU does at 5.9 GHz with higher voltage. It is a great CPU and I am happy with it. But, I can hardly wait to get it on bare die. It's just way too hot with the factory solder unless I run it on chilled water. 

 

I think I am going to put one of my systems back on the open bench again. I'm getting tire of the vertical motherboard arrangement, so I will probably sell the be quiet! Dark Base Pro 901 and just live with the dust. I did get an air purifier about a week ago and it has helped a great deal with the dust in my office. I do love the Dark Base Pro 901, but the horizontal orientation of motherboards is just way superior compared to a vertical orientation. Every time I see the Praxis Wetbench in my closet I wish I was still using it. I'm so glad I did not sell it like I was thinking I might.

 

Yep I always ran 8400/8533 with tight timings it performs really good. But the main reason why was because my prior chip(s) could not run 8600 with tight timings. This chip will be worked up to a 8600 tight profile. 
 

As for moving away from direct die. I just don’t need it. Liquid Metal on 15C water daily doesn’t last very long. So it requires maintenance every 1-3 months. 
 

I can run either of these 14900KS chips at 5.9Ghz(8P) 6.2Ghz (2P) with a load voltage of (1.190 higher SP chip) OR (1.208V lower SP chip) The package temps are substantially warmer than direct die. But the CPU is still only 67-68-69C like this in Cinebench. The solder and my mount is good for a very very long time. Chiller and Liquid Metal daily usage breaks down and etches badly and doesn’t hold up worth a dang. I have always had this problem, and I’m just tired of dealing with it. 

[Mr. Fox]

1 hour ago, tps3443 said:

 

Yep I always ran 8400/8533 with tight timings it performs really good. But the main reason why was because my prior chip(s) could not run 8600 with tight timings. This chip will be worked up to a 8600 tight profile. 
 

As for moving away from direct die. I just don’t need it. Liquid Metal on 15C water daily doesn’t last very long. So it requires maintenance every 1-3 months. 
 

I can run either of these 14900KS chips at 5.9Ghz(8P) 6.2Ghz (2P) with a load voltage of (1.190 higher SP chip) OR (1.208V lower SP chip) The package temps are substantially warmer than direct die. But the CPU is still only 67-68-69C like this in Cinebench. The solder and my mount is good for a very very long time. Chiller and Liquid Metal daily usage breaks down and etches badly and doesn’t hold up worth a dang. I have always had this problem, and I’m just tired of dealing with it. 

That makes sense. And I suspect I haven't had the issues since I don't run the chiller all the time. Have you tried running a really good thermal paste on bare die instead of liquid metal? I suspect that would hold up better. I had to on the 5950X because of the chiplets. Liquid metal worked poorly and was worse than stock solder because the chiplets had a very minor z-height difference between them and liquid metal did not make good enough contact. The temps with liquid metal were worse than thermal paste due to poor contact and thermal paste was identical to stock solder, so the entire thing was a monumental waste of time and money.

 

If I were still using a Supercool direct die setup like yours that would be a deal-ender having to reapply the liquid metal that often. The IceMan direct die is far less cumbersome and no different than removing an ordinary water block. No need to break out paper towel and worry about leaking o-rings, etc. The extra effort required was why I stopped using it.

 

Just for giggles I should try thermal paste with the IceMan to see how it works. I have never used anything except liquid metal.

 

[tps3443]

1 hour ago, Mr. Fox said:

That makes sense. And I suspect I haven't had the issues since I don't run the chiller all the time. Have you tried running a really good thermal paste on bare die instead of liquid metal? I suspect that would hold up better. I had to on the 5950X because of the chiplets. Liquid metal worked poorly and was worse than stock solder because the chiplets had a very minor z-height difference between them and liquid metal did not make good enough contact. The temps with liquid metal were worse than thermal paste due to poor contact and thermal paste was identical to stock solder, so the entire thing was a monumental waste of time and money.

 

If I were still using a Supercool direct die setup like yours that would be a deal-ender having to reapply the liquid metal that often. The IceMan direct die is far less cumbersome and no different than removing an ordinary water block. No need to break out paper towel and worry about leaking o-rings, etc. The extra effort required was why I stopped using it.

 

Just for giggles I should try thermal paste with the IceMan to see how it works. I have never used anything except liquid metal.

 

 

Yeah the maintenance is higher on Supercool or any direct die with daily chiller+LM. But its temps are just insanely good no doubt. 5.9Ghz would be 50C cpu package temps with my supercool in R23 using direct die. I considered thermal paste on direct die, but now I felt I’m going backwards now. Stock solder is supposed to be better than thermal paste 😁. The only thing direct die and Liquid Metal will give me is a 15C cooler CPU. And +100Mhz.

 

I can run either of these KS chips, and my R-Batch at 6.0P/4.8E with no delid, and still stay in the 70’s on CPU package. I think that’s pretty good. 

[Raiderman]

Congratulations bro @chew 🤙🤙

 

[Mr. Fox]

11 hours ago, tps3443 said:

 I considered thermal paste on direct die, but now I felt I’m going backwards now. Stock solder is supposed to be better than thermal paste 😁. The only thing direct die and Liquid Metal will give me is a 15C cooler CPU. And +100Mhz.

I thought that way initially, but then logic kicked in and I realized it just can't be true. How is using thermal paste with direct die not better than solder+IHS+thermal paste? You're still using thermal paste, but with extra crap in between to trap the heat. The closer you can get the cooling plate to the die the better your temps will be unless something does not fit correctly. Imagine how horrible laptop temperatures would be if they were not direct die. They'd be totally unusable. So, you would be going backwards a little bit, but it would/should still be measurably better than with the IHS.

 

Maybe when I delid this 14900KF I received yesterday I will compare stock with thermal paste, direct die with thermal paste, and then direct die with liquid metal with no chiller and see how those numbers look.

 

Imagine how much worse your temps would be without the chiller. The chiller makes up for a lot of thermal management problems on a CPU that is not delidded and direct die. You'd be down probably 300 MHz and about 30°C hotter if you didn't drop 200-300 MHz to cope with it. In fact, your CPU probably would not be usable at the clock speeds you are running now with the chiller if you did not have the chiller. You'd be limited to 5.6 or 5.7 GHz tops and that would be hitting above 100°C. And, the only reason I know this is because I only run the chiller for benching. The main difference is my ambient temperatures are probably 10°-15°F higher than yours on my hottest summer days.

[electrosoft]

16 hours ago, Raiderman said:

Vulkan doesn't run as well as dx12 on Amd. At least that particular bench doesn't.

 

I'm curious how your system would do with that bench?

 

I'll have to give it a run once it's up and in running condition. Bad argb hub and bad PSU stalled the build.

 

 

[electrosoft]

14 minutes ago, Mr. Fox said:

Maybe when I delid this 14900KF I received yesterday I will compare stock with thermal paste, direct die with thermal paste, and then direct die with liquid metal with no chiller and see how those numbers look.

 

 

This would be a nice comparison. With and without chiller too would be nice to have a complete set of data in both states.

 

 

[tps3443]

49 minutes ago, Mr. Fox said:

I thought that way initially, but then logic kicked in and I realized it just can't be true. How is using thermal paste with direct die not better than solder+IHS+thermal paste? You're still using thermal paste, but with extra crap in between to trap the heat. The closer you can get the cooling plate to the die the better your temps with be unless something does not fit correctly. Imagine how horrible laptop temperatures would be if they were not direct die. They'd be totally unusable. So, you would be going backwards a little bit, but it would/should still be measurable better than with the IHS.

 

Maybe when I delid this 14900KF I received yesterday I will compare stock with thermal paste, direct die with thermal paste, and then direct die with liquid metal with no chiller and see how those numbers look.

 

Imagine how much worse your temps would be without the chiller. The chiller makes up for a lot of themal management problems on a CPU that is not delidded and direct die. You'd be down probably 300 MHz and about 30°C hotter if you didn't drop 200-300 MHz to cope with it. In fact, your CPU probably would not be usable at the clock speeds you are running now with the chiller if you did not have the chiller. You'd be limited to 5.6 or 5.7 GHz tops and that would be hitting above 100°C. And, the only reason I know this is because I only run the chiller for benching. The main difference is my ambient temperatures are probably 10°-15°F higher than yours on my hottest summer days.

This chiller helps a lot but not nearly as much as direct die does. With a rad only, my cpu would only be 10c warmer. So I could still run 5.9/4.8 easily and be 78-79C package on normal ambient water without a delid or direct die. I tested this on my R-Batch, and this SP108 has a really low 5.9 Vmin for a soldered chip, lower than my R-batch when it was soldered. You might even be able to hulk it to 6Ghz if you don’t mind 89-90C temps and higher voltage lol.

[Mr. Fox]

1 hour ago, tps3443 said:

This chiller helps a lot but not nearly as much as direct die does. With a rad only, my cpu would only be 10c warmer. So I could still run 5.9/4.8 easily and be 78-79C package on normal ambient water without a delid or direct die. I tested this on my R-Batch, and this SP108 has a really low 5.9 Vmin for a soldered chip, lower than my R-batch when it was soldered. You might even be able to hulk it to 6Ghz if you don’t mind 89-90C temps and higher voltage lol.

For me without the chiller 5.8 with 1.250V under load in Cinebench is 100-105°C with ambient temperaturs of 80°F without the chiller. Gaming is fine that way for sure, nowhere even close to those temperatures, but not benching the CPU.

[Papusan]

3 hours ago, tps3443 said:

 

Yeah the maintenance is higher on Supercool or any direct die with daily chiller+LM. But its temps are just insanely good no doubt. 5.9Ghz would be 50C cpu package temps with my supercool in R23 using direct die. I considered thermal paste on direct die, but now I felt I’m going backwards now. Stock solder is supposed to be better than thermal paste 😁. The only thing direct die and Liquid Metal will give me is a 15C cooler CPU. And +100Mhz.

 

I can run either of these KS chips, and my R-Batch at 6.0P/4.8E with no delid, and still stay in the 70’s on CPU package. I think that’s pretty good. 

 

What brand and type Liquid metal have you used? Maybe change brand and type Liquid metal. They are mixed differently so maybe you could try that.

 

@Mr. FoxThe way all Windows versions should have worked. Optional options. Microsoft force the gamer-kids and aunt mary Secure boot and whatever. But corps that need more security can do whatever they want. Disgusting?

 

 

Windows 11 LTSC 2024 arrives making TPM and Secure Boot optional — lower storage requirements, too

[Reciever]

1 minute ago, Papusan said:

 

What brand and type Liquid metal have you used? Maybe change brand and type Liquid metal. They are mixed differently so maybe you could try that.

 

@Mr. FoxThe way all Windows versions should have worked. Optional options. Microsoft force the gamer-kids and aunt mary Secure boot and whatever. But corps that need more security can do whatever they want. Disgusting?

 

Windows 11 LTSC 2024 arrives making TPM and Secure Boot optional — lower storage requirements, too

 

 

The sluggish response of the forum has been observed on my end as well, will reach out to Hiew when he gets back from visiting his family. 

 

Also that link should go in the Win11 thread 

[Papusan]

3 minutes ago, Reciever said:

 

 

The sluggish response of the forum has been observed on my end as well, will reach out to Hiew when he gets back from visiting his family. 

 

Also that link should go in the Win11 thread 

 

Thanks 🙂

 

This is the benchmark thread. A stripped down OS coming from Microsoft should also apply in this thread 🙂 What OS you use is important for benching. I can also put it in the Windows 11 thread.

[Reciever]

2 minutes ago, Papusan said:

 

Thanks 🙂

 

This is the benchmark thread. A stripped down OS coming from Microsoft should also apply in this thread 🙂 What OS you use is important for benching. I can also put it in the Windows 11 thread.

That would be crossposting, which is explicitly not endorsed on the forum :)

 

If you had benched something using that version of OS, it would be more than fine. However there is no mention of benchmarking what so ever, instead pointing to IoT devices. We play it pretty loose here but thats reaching. 

[Talon]

On 5/23/2024 at 8:51 PM, tps3443 said:

I am doing a little 14900KS binning for fun. I’ve got to say though, I’m impressed by the SP99 KS chip lol. Insanely good IMC its an MCSP 82, very strong P Cores with 5.9+ of course, It could not touch my R batch which to this day will still do (5.7Ghz (8P) (6.0Ghz 2P)/4.4E XMP 8200@1.110V load (Even while re-lidded) and 238 watts R23.

 

Anyways, I have another 14900KS that will be here tomorrow. And if it has a weak IMC, I’m going to proudly keep the SP99 chip which I just can’t fault for a second it’s a beast of a performer. I was so impressed that the SP99 could run DDR5 8400 (262k tREFI) with less than 1.400V VDD/VDDQ. And also hold down 5.9P-6.2P@1.208V load.
 

I’m okay if the chip doesn’t have ground breaking P/E cores. 
 

I can’t really recommend a 14900KS if someone has an AIO cooler though. Maybe if it’s some crazy good chip. The thing is, a 14900K or 14900Kf will more than likely run 5.7 or even 5.8Ghz with less load voltage/load power and less load heat. So this makes a KS completely useless with weak cooling since they are literally binned to run for 5.9+ ONLY. Hence why you have all of these bad reviews of people hating on them saying it’s worse than their K/KF.  😎 

 

I’m wondering how many returned KS chips there are that were actually probably strong lol.

 

I sold an SP99 13900K I had in the closet this morning. I put it in my system last night to ensure it was the SP99 chip, and did a quick test on it. 

 

It scored 43K in CB23? I ran this test multiple times and same results, highest real-time was 43.6K. I was like wtf. I've never scored that high at these power limits with a 420mm AIO and just 5.5Ghz P core stock. Is there some sort of Windows OS cancer that is getting applied to these Intel CPUs that didn't apply to mine? I then tested TimeSpy and score over 27K CPU, again abnormally high for a stock 13900K. 

 

I rebooted and checked power limits and mess with undervolt a tad, and upon reboot the score dropped to 40.6K. I was never able to get the score back, even with undervolt, overclock or power limit changes. 

 

I am now fully convinced Windows is applying some sort of cancer patch to these CPUs and harming performance. Have any of you seen this phenomena? All I did was swap my 14900KS to the 13900K. I sloppily attached the cooler and barely torqued it down as I figured I was just going to verify the SP and pull it. 

 

[tps3443]

37 minutes ago, Papusan said:

 

What brand and type Liquid metal have you used? Maybe change brand and type Liquid metal. They are mixed differently so maybe you could try that.

 

@Mr. FoxThe way all Windows versions should have worked. Optional options. Microsoft force the gamer-kids and aunt mary Secure boot and whatever. But corps that need more security can do whatever they want. Disgusting?

 

 

Windows 11 LTSC 2024 arrives making TPM and Secure Boot optional — lower storage requirements, too

I tried a few. Thermalright, Bequiet, thermal grizzly. They can all last months and months, and after like 3-4 months the performance just fades a little bit but still okay. But when it eventually fails you lose that solid stability, and that day will always be a surprise visit out of the blue lol. The failure can be closing to desktop when gaming, or other random stability issues. Then you have to pull the chip and prep everything again with some Flitz and possibly sand paper if it’s really bad. This was one of the main reasons for selling my R-Batch. Now, I think Liquid Metal and ambient water is totally fine. But with 15C water it just fails really fast. 

 

6 minutes ago, Talon said:

 

I sold an SP99 13900K I had in the closet this morning. I put it in my system last night to ensure it was the SP99 chip, and did a quick test on it. 

 

It scored 43K in CB23? I ran this test multiple times and same results, highest real-time was 43.6K. I was like wtf. I've never scored that high at these power limits with a 420mm AIO and just 5.5Ghz P core stock. Is there some sort of Windows OS cancer that is getting applied to these Intel CPUs that didn't apply to mine? I then tested TimeSpy and score over 27K CPU, again abnormally high for a stock 13900K. 

 

I rebooted and checked power limits and mess with undervolt a tad, and upon reboot the score dropped to 40.6K. I was never able to get the score back, even with undervolt, overclock or power limit changes. 

 

I am now fully convinced Windows is applying some sort of cancer patch to these CPUs and harming performance. Have any of you seen this phenomena? All I did was swap my 14900KS to the 13900K. I sloppily attached the cooler and barely torqued it down as I figured I was just going to verify the SP and pull it. 

 

 

Never seen a 13900K do that well before! Thats faster than a stock 14900KS  😂 

 

 

[Mr. Fox]

3 hours ago, Talon said:

 

I sold an SP99 13900K I had in the closet this morning. I put it in my system last night to ensure it was the SP99 chip, and did a quick test on it. 

 

It scored 43K in CB23? I ran this test multiple times and same results, highest real-time was 43.6K. I was like wtf. I've never scored that high at these power limits with a 420mm AIO and just 5.5Ghz P core stock. Is there some sort of Windows OS cancer that is getting applied to these Intel CPUs that didn't apply to mine? I then tested TimeSpy and score over 27K CPU, again abnormally high for a stock 13900K. 

 

I rebooted and checked power limits and mess with undervolt a tad, and upon reboot the score dropped to 40.6K. I was never able to get the score back, even with undervolt, overclock or power limit changes. 

 

I am now fully convinced Windows is applying some sort of cancer patch to these CPUs and harming performance. Have any of you seen this phenomena? All I did was swap my 14900KS to the 13900K. I sloppily attached the cooler and barely torqued it down as I figured I was just going to verify the SP and pull it. 

 

Run at the same clock speeds, my 13900KS outperforms all of my 14900KF by between 500 and 1000 Cinebench R23 points depending on what kind of Windoze trash OS it is run on. Probably due to more security mitigation filth baked into the newer processors. That seldom has desirable outcomes.

 

Edit: @Talon if you have the Park Control app, launch it and see how many of your cores are being forced to park. If you are running the latest cancer versions of W10/11, I was experiencing something similar and even using the Park Control utility. Cinebench scores were down. I had to fart around a bit to get Winduhz 11 to leave my CPU alone and stop screwing with it to save power. Really made me mad, so I wiped the drive and downgraded to 22H2 and everything was as it should be, and updates are blocked so it can't poop on it again later. I don't think the asshats responsible for Windoze development could do anything right if their lives depended on it.

 

 

Edited May 27 by Mr. Fox
added info

[Papusan]

3 hours ago, Mr. Fox said:

Edit: @Talon if you have the Park Control app, launch it and see how many of your cores are being forced to park. If you are running the latest cancer versions of W10/11, I was experiencing something similar and even using the Park Control utility. Cinebench scores were down. I had to fart around a bit to get Winduhz 11 to leave my CPU alone and stop screwing with it to save power. Really made me mad, so I wiped the drive and downgraded to 22H2 and everything was as it should be, and updates are blocked so it can't poop on it again later. I don't think the asshats responsible for Windoze development could do anything right if their lives depended on it.

 

Expect Microsoft will continue test and change power profiles in windows to fit their new low power ARM processors. It's all about power saving and how long the new junk can run on the battery with a YouTube video in the background. It's all about maxing out battery life. And desktop hardware will get exactly same power settings. Only with minor changes to scheduler for the Intel hybrid Junk.

 

Microsoft announces new Surface Laptop with Snapdragon processors

[Mr. Fox]

3 minutes ago, Papusan said:

 

Expect Microsoft will continue test and change power profiles in windows to fit their new low power ARM processors. It's all about power saving and how long the new junk can run on the battery with a YouTube video in the background. It's all about maxing out battery life. And desktop hardware will get exactly same power settings. Only with minor changes to scheduler for the Intel hybrid Junk.

 

Microsoft announces new Surface Laptop with Snapdragon processors

Sometimes images are more accurate than words, and here is what I think about that. And, them.