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

[Papusan]

35 minutes ago, electrosoft said:

It could have been worse 😞 

 

Yup, it could be worse. Much worse. What with SP94?🤮 The AI binning have gone all too far. This KS launch looks worse and worse for every new day. Sum up... Disgusting

 

[Mr. Fox]

2 hours ago, Rage Set said:

Well, my 14900KS is only a 103SP. My poor luck with Intel chips continues. Since I used some of my CC rewards to pay for it, I'm not sure if I want to go through the process of returning it. Essentially, I only paid 610 shipped. 

1 hour ago, electrosoft said:

It could have been worse 😞 Make sure to test it especially the IMC if you're looking to push 8400+

Yes, it could have been worse indeed. Brother @Rage Set what does the VF curve look like? If you haven't already done so, you can reflash to BIOS to an older verion, then back up to the version you intend to use. Sometimes the SP ratings get corrupted or reported inaccurately. But the VF curve should be accurate and consistent. The SP ratings are often wrong after installing a different CPU and sometimes after removing and reinstalling the same CPU. 

1 hour ago, Papusan said:

 

Yup, it could be worse. Much worse. What with SP94?🤮 The AI binning have gone all too far. This KS launch looks worse and worse for every new day. Sum up... Disgusting

 

I could not understand what was being said, but the reason for the laughter seems obvious. Although crying might be more appropriate than laughing.

[Rage Set]

56 minutes ago, Mr. Fox said:

Yes, it could have been worse indeed. Brother @Rage Set what does the VF curve look like? If you haven't already done so, you can reflash to BIOS to an older verion, then back up to the version you intend to use. Sometimes the SP ratings get corrupted or reported inaccurately. But the VF curve should be accurate and consistent. The SP ratings are often wrong after installing a different CPU and sometimes after removing and reinstalling the same CPU. 

I could not understand what was being said, but the reason for the laughter seems obvious. Although crying might be more appropriate than laughing.

 

Based off what I am seeing, I won't be touching 8400MT with this chip, hahaha. My MCSP is 79. Sorry for the lowres, I am using my EVGA 570 for the output. 

 

 

[Mr. Fox]

46 minutes ago, Rage Set said:

 

Based off what I am seeing, I won't be touching 8400MT with this chip, hahaha. My MCSP is 79. Sorry for the lowres, I am using my EVGA 570 for the output. 

 

 

Looks like it was a near twin to the turd chip I returned. SP102 (P117/E77) and here is the VF curve on that one... The cooler rating will directly affect the voltage values showing below the SP rating number and so will the LLC level, but those won't affect the VF curve.

 

---

 

I cannot remember now from my X570 Crosshair @Rage Set maybe you or @jaybee83 know, does ASUS have an Silicon Prediction (SP) rating system for the AMD motherboards? 

[Rage Set]

48 minutes ago, Mr. Fox said:

Looks like it was a near twin to the turd chip I returned. SP102 (P117/E77) and here is the VF curve on that one... The cooler rating will directly affect the voltage values showing below the SP rating number and so will the LLC level, but those won't affect the VF curve.

 

---

 

I cannot remember now from my X570 Crosshair @Rage Set maybe you or @jaybee83 know, does ASUS have an Silicon Prediction (SP) rating system for the AMD motherboards? 

 

No. AMD technically has their own, plain vanilla, SP rating system built into Ryzen Master. However, 1usmus developed Clock Tuner that sat ontop of Ryzen Master that gave you an "SP" rating. Through that software, I discovered I had a lot of winners, especially my former 5800X. 

 

1usmus stopped developing it, around the time the rumors of AM5 was set to launch. 

[Mr. Fox]

1 hour ago, Rage Set said:

 

No. AMD technically has their own, plain vanilla, SP rating system built into Ryzen Master. However, 1usmus developed Clock Tuner that sat ontop of Ryzen Master that gave you an "SP" rating. Through that software, I discovered I had a lot of winners, especially my former 5800X. 

 

1usmus stopped developing it, around the time the rumors of AM5 was set to launch. 

Too bad he did not continue development. I would trust a third-party development like that more than one developed by the ODM, since the third-party developer would have no reason to care whether the end-user had a good CPU or trash sample.

[Papusan]

2 hours ago, Rage Set said:

 

Based off what I am seeing, I won't be touching 8400MT with this chip, hahaha. My MCSP is 79. Sorry for the lowres, I am using my EVGA 570 for the output. 

 

 

 

Disgusting VF curve. My worse SP 100 defective chips have better VF curve for 5100, 5600 and 5900. But your 6200 is massive 5mv better. Great value for 3 points better chips. I expect that's matter most for Intel when they bin for the 6.2GHz Special Edition😕 

 

 

And here is 6 new samples in a row of defective KS chips from Norway. From SP99 to 102. Nice.

https://www.overclock.net/threads/overclocking-raptor-lake-refresh-14900k-14700k-14600k-etc-results-bins-and-discussion.1807439/post-29310183

[electrosoft]

2 hours ago, Mr. Fox said:

Looks like it was a near twin to the turd chip I returned. SP102 (P117/E77) and here is the VF curve on that one... The cooler rating will directly affect the voltage values showing below the SP rating number and so will the LLC level, but those won't affect the VF curve.

 

---

 

I cannot remember now from my X570 Crosshair @Rage Set maybe you or @jaybee83 know, does ASUS have an Silicon Prediction (SP) rating system for the AMD motherboards? 

 

Asus now has SP ratings for AMD chips for AM5 for their upper tier boards and will give an SP rating per each CCD and a combined score which is kind of cool along with V/F curves.

 

For AMD, 7800X3D / CCDx X3D on a 7950X3D, anything over SP100 is above average. SP103+ is golden.

 

CCDx non X3D, anything above ~117 is above average and over 120 is golden.

 

Combined, anything greater than 112 is golden.

 

Example:

 

 

 

[Raiderman]

1 hour ago, electrosoft said:

 

Asus now has SP ratings for AMD chips for AM5 for their upper tier boards and will give an SP rating per each CCD and a combined score which is kind of cool along with V/F curves.

 

For AMD, 7800X3D / CCDx X3D on a 7950X3D, anything over SP100 is above average. SP103+ is golden.

 

CCDx non X3D, anything above ~117 is above average and over 120 is golden.

 

Combined, anything greater than 112 is golden.

 

Example:

 

 

 

I'd like to see MSI develop something along those lines. By the time they do, I'll have moved on from them to AsRock.

[tps3443]

5 hours ago, Rage Set said:

 

Based off what I am seeing, I won't be touching 8400MT with this chip, hahaha. My MCSP is 79. Sorry for the lowres, I am using my EVGA 570 for the output. 

 

 

I think it might just surprise you! Test it out and see what it clocks to. Also, when you are running stability testing make sure you try tighter timings, it seems some chips will fail stability with loose timings, but easily pass with tighter timings same frequency. 
 

My 14900KF is like this. I can load and XMP profile of 8400c40-52-52-52-131 and it fails TM5 in 12 seconds or less, bump ran voltage to 1.500V and it still fails, then I can drop it to 8400 C36-48-48-48-60 and pass anything under the sun with very minimal voltages. It is an odd thing. But was an interesting moment to learn that some chips just LOVE tight timings and won’t work with loose timings. 

[Papusan]

Intel bashed AMD for their use of glued dies. And soon we'll see Intel going even more retarded but with glued dummy dies paired with multiple glued on tiles, LOOL

Intel Arrow Lake-H processor with 24 cores has reportedly been photographed
https://videocardz.com/newz/intel-arrow-lake-h-processor-with-24-cores-has-reportedly-been-photographed

 

[electrosoft]

2 hours ago, Raiderman said:

I'd like to see MSI develop something along those lines. By the time they do, I'll have moved on from them to AsRock.

 

Yeah, I've got my eye on Asrock. They left a sour taste in my mouth with the Z390 Taichi but they have slowly redeemed themselves through a couple of MBs and GPUs over the years since.

 

White Asrock Taichi 7900xtx is down to $999.99. If I hadn't snagged this Hellhound 7900xtx for $720 total to my door, I prolly would have picked up an open box for $860+tax = ~$914 shipped.

 

1 hour ago, tps3443 said:

I think it might just surprise you! Test it out and see what it clocks to. Also, when you are running stability testing make sure you try tighter timings, it seems some chips will fail stability with loose timings, but easily pass with tighter timings same frequency. 
 

My 14900KF is like this. I can load and XMP profile of 8400c40-52-52-52-131 and it fails TM5 in 12 seconds or less, bump ran voltage to 1.500V and it still fails, then I can drop it to 8400 C36-48-48-48-60 and pass anything under the sun with very minimal voltages. It is an odd thing. But was an interesting moment to learn that some chips just LOVE tight timings and won’t work with loose timings. 

 

Yep. When I was first testing 8000 on my 13900KS, it bombed spectacularly just scaling at Jedec 4800 loose timings . XMP eventually bombed out, but once I tightened them up they had no problem and that's how I ran them.

 

 

 

 

[chew]

 

 

moving on to 8700g class

 

[tps3443]

Anyone know when MSI is planning on dropping the new Mpower Z790 two dimm boards for $199.99? I have not heard anything. Are they in other countries? 

[Mr. Fox]

6 minutes ago, tps3443 said:

Anyone know when MSI is planning on dropping the new Mpower Z790 two dimm boards for $199.99? I have not heard anything. Are they in other countries? 

I wouldn't mind getting my hands on one. From what I read they are not going to release them in European or American markets, only Asian distributions. If that is true it really sucks.

[Talon]

22 hours ago, Rage Set said:

 

Based off what I am seeing, I won't be touching 8400MT with this chip, hahaha. My MCSP is 79. Sorry for the lowres, I am using my EVGA 570 for the output. 

 

 

 

Don't pay any attention to that MC score. It's total and complete nonsense. If you have a good 2 dimmer board, I have yet to find a single 14900KS that can't do 8200+. That score is with your memory at autos. Set XMP and check SP again, I bet it's over 90. I've discovered the IMC score changes a lot, it's not reliable. 

 

https://imgur.com/kh6qRjw

 

Here is my 14900KS SP109 with running XMP II 8200 CL38 with just 1.225v IMC !! Asus auto settings were 1.4v to the IMC. I've been shaving it down the last couple days. BF2042 and memory tests and I still have yet to crash the system. I'm going to keep shaving down. These IMCs seem rock solid on 14900KS, at least some of them. 

[Raiderman]

19 hours ago, electrosoft said:

 

Yeah, I've got my eye on Asrock. They left a sour taste in my mouth with the Z390 Taichi but they have slowly redeemed themselves through a couple of MBs and GPUs over the years since.

 

White Asrock Taichi 7900xtx is down to $999.99. If I hadn't snagged this Hellhound 7900xtx for $720 total to my door, I prolly would have picked up an open box for $860+tax = ~$914 shipped.

 

 

Yep. When I was first testing 8000 on my 13900KS, it bombed spectacularly just scaling at Jedec 4800 loose timings . XMP eventually bombed out, but once I tightened them up they had no problem and that's how I ran them.

 

 

 

 

That X570 Asrock board I had was rock solid. Same as the 7900xtx I have now. They have come a long way over the last few years. I guess it's the same with any manufacturer, they all have ups and downs.

[win32asmguy]

I finally have some parts collected to build a desktop system to use as a media encoder and server. I can put the delidded 12900KS to use that I have had over a year now from the NH55 sitting in a box. Got a MSI Z690-A, 32GB DDR5 6400, Noctua D15 and a 2TB WD850 that I will use as a boot and scratch drive. I still need to figure out what I want to do for media storage drives, not sure what is the best for that use case these day. This system will have to run Windows 10 so far as I can tell as Makemkv seems to not support Linux.

 

Hopefully it will not be very noisy while encoding as it will sit near our TV and couch. Right now I have an external laptop bluray drive but hope to get a faster internal drive so long as its somewhat quiet while ripping. I will note it is also hard these days to find a nice case that still has a 5.25 drive slot!

[Raiderman]

3 hours ago, win32asmguy said:

I finally have some parts collected to build a desktop system to use as a media encoder and server. I can put the delidded 12900KS to use that I have had over a year now from the NH55 sitting in a box. Got a MSI Z690-A, 32GB DDR5 6400, Noctua D15 and a 2TB WD850 that I will use as a boot and scratch drive. I still need to figure out what I want to do for media storage drives, not sure what is the best for that use case these day. This system will have to run Windows 10 so far as I can tell as Makemkv seems to not support Linux.

 

Hopefully it will not be very noisy while encoding as it will sit near our TV and couch. Right now I have an external laptop bluray drive but hope to get a faster internal drive so long as its somewhat quiet while ripping. I will note it is also hard these days to find a nice case that still has a 5.25 drive slot!

Traditional hard disk drives are still the most economical way to store large files. If you need faster access than that, you will spend $$ of course. You could always stripe a couple together in a Raid 0 array, but redundancy goes out the window. Maybe some 2.5 SSD's?

[chew]

this lemons dry now......

 

AMD Radeon 780M (desktop) video card benchmark result - AMD Ryzen 7 8700G,Gigabyte Technology Co., Ltd. B650E AORUS TACHYON (3dmark.com)

 

[tps3443]

9 hours ago, Talon said:

 

Don't pay any attention to that MC score. It's total and complete nonsense. If you have a good 2 dimmer board, I have yet to find a single 14900KS that can't do 8200+. That score is with your memory at autos. Set XMP and check SP again, I bet it's over 90. I've discovered the IMC score changes a lot, it's not reliable. 

 

https://imgur.com/kh6qRjw

 

Here is my 14900KS SP109 with running XMP II 8200 CL38 with just 1.225v IMC !! Asus auto settings were 1.4v to the IMC. I've been shaving it down the last couple days. BF2042 and memory tests and I still have yet to crash the system. I'm going to keep shaving down. These IMCs seem rock solid on 14900KS, at least some of them. 

The most useful thing with mcsp is merely for using it as our own tool to compare with ourselves. Checking mcsp after swapping to a new bios version. Usually a lower mcsp will result in worse ram OC. So, if I have MCSP 90-92 on bios 1501, but suddenly have MCSP 87 or 88 on bios 1901. This will usually make edge of ram OC stability much much harder for me. This has been a pretty reliable tool for so far. 
 

What is your MCSP? Usually high mcsp can run low IMC VDD. Also, your MCSP jumps that much by setting just XMP? It goes up 11 points? Holy cow man. lol. 😂 

 

[Papusan]

44 minutes ago, tps3443 said:

The most useful thing with mcsp is merely for using it as our own tool to compare with ourselves. Checking mcsp after swapping to a new bios version. Usually a lower mcsp will result in worse ram OC. So, if I have MCSP 90-92 on bios 1501, but suddenly have MCSP 87 or 88 on bios 1901. This will usually make edge of ram OC stability much much harder for me. This has been a pretty reliable tool for so far. 
 

What is your MCSP? Usually high mcsp can run low IMC VDD. Also, your MCSP jumps that much by setting just XMP? It goes up 11 points? Holy cow man. lol. 😂 

 

 

Fun part. Reset bios and run mcsp again and you may get different results 🙂 No need for a different bios to get different mcsp results. Tested this yesterday before I removed the defective KS for return to sender. The mcsp went from 80 to 77. Is less more?😀 And the defective KS couldn't run higher than 47 on the E-cores. In short... Intel didn't bother with the E-cores when they binned for the 62x special Edition. They totally forgot that their defective junk have hybrid architecture.

 

 

[Mr. Fox]

Mine changes often as well. It is often not the same between BIOS version changes on either of the Apex. It can vary anywhere from 1 to 5 points up or down. It is often 1 or 2 points lower the second time if I run it twice in a row.

[tps3443]

24 minutes ago, Papusan said:

 

Fun part. Reset bios and run mcsp again and you may get different results 🙂 No need for a different bios to get different mcsp results. Tested this yesterday before I removed the defective KS for return to sender. The mcsp went from 80 to 77. Is less more?😀 And the defective KS couldn't run higher than 47 on the E-cores. 

 

 

Yes but that’s probably just normal deviation of like 3%. Going from 79 to 90+ would be pretty astounding though. @Rage Set enable XMp and test again. Also, install bios 2102. After installing this latest bios it bumped my mcsp to 89-90 with a cleared CMOS, and mcsp 91-92 with XMP enabled. 
 

Testing 2102 has shown good ram OC improvements. Stabilizing 8400c36 was very quick. 
 

PS: My mcsp has never hit 90 since owning this chip. Always 88-89 on every bios I have tried. Once installing 2102 I saw it hit mcsp 91 then mcsp 92. This is always a good sign. And I can tell right away with ram overclocking during the stability testing if it’s easier or better. I’ve been ram overclocking on this Z790 Apex for about a year now, and testing this same chip for almost 4 months. I’ve tried 1402/1501/1801/1904/9901/2102. Of all of these 1501 overclocked ram the best, that is until 2102 came along. 😁

[Reciever]

So @Hiew ran an idea by me last night and unfortunately the proposition does seem quite attractive given the price points things are right now.

 

Heres the scenario I have been entertaining, Selling the RTX 3090 for about 800(+/-) and buying the 7900 XTX. The XTX looks to be a healthy bump in performance over the 3090 and in some titles actually quite substantial. Would only need to spend about 150 or so tops for the upgrade.

 

Before anyone points it out I really dont care about Ray Tracing or DLSS. Raster performance is more important to me.