Swap 13700k for 7800X3D

[warriorunited]

Yes lol. he means 50mV, or 0.050V.

I did that, and the result was: Cinebench instant crash after 5 seconds lol

 

[Bobaganoosh]

I did that, and the result was: Cinebench instant crash after 5 seconds lol

interesting. Well...after you changed the Asus performance boost setting, it seemed like your voltages were much better anyway. You could try like a 0.030 for a slight offset, but your current numbers look like they're much more reasonable than before and it really shouldn't be as hot as it is. I'd start verifying the mount. Do you have any newer thermal paste around? Kryosheets are handy for ease of use (38x38 for intel 1700), but Noctua NT-H2, be quiet! DC2, Prolimatech PK-3 Nano are good reasonably priced pastes lately. PTM7950 phase change sheets have been getting some good reviews lately too.

 

[oxrufiioxo]

First thing I'd do is get a contact frame and maybe for ease of use get a kryosheet that fits your processor 38x38.

If that doesn't work

The second thing I'd do is from a seller with an excellent return policy buy a known good cooler just for testing.

If that doesn't work

The last thing I'd do also fron a seller with a good return policy is buy a 13700k and test to make sure both cpus behave the same.

Maybe none of those steps is an option for you but I'd at least get a contact frame.

Obviously after you've dialed in the bios which honestly a user shouldn't even have to do just to get their cpu to not hit 100c. That being said it really comes with the territory of owning a high end 13th/14th generation part they've always shipped with dubious out of box settings. Ryzen X parts aren't much different they all hit 95c and the X3D parts around 85c.... They just aren't degrading themselves at least as rapidly.

Also unless someone is using your exact cooling/case/motherboard/cpu/ambient comparing temps is mostly useless you could've just lost the lottery on both Cpu quality and the IHS ability to transfer heat if you purchased 10 13700ks they'd all run a bit differently throw in the fact that some degrade themselves and it's really a crap shoot trying to compare with anyone else's numbers.

That being said I wouldn't jump ship at least not right now I'd wait till AMD/Intel offers you a meaningful upgrade while ditching this troublesome platform assuming none of the above steps fixes the issue in a way that is satisfactory to you.

 

[Bill_Bright]

The lack of a symptom does not preclude having a disease that's known to be a asmptomatic until a certain point.

What a silly, nonsensical comment! Are you even listening to yourself?

evernessince to the doctor, "Hey doc! I have no symptoms but, since humans get cancer I might have it one day so you better start chemo now and take out my colon too!"

Or, "Hey Corsair! My PSU is super quiet but since some Corsairs have been reported to have coil whine, mine might one day too. So give me a new one now."

there will be hundreds of thousands of people with Intel CPUs that have suffered silent degradation

You don't know that. This is you jumping to unfounded conclusions again. It seems you have convinced yourself that everyone who bought one of those Intel processor models not only is affected by the micro-code bug, but their CPU has suffered damage from it.

I have to wonder about the potential of bias here, noting the AMD in your system specs.

MAKE NO MISTAKE - I am NOT defending Intel in any way here. I am simply pointing out the common sense aspect. Unless or until there is a total recall of your specific model number, you cannot simply call up a manufacturer and demand a replacement or refund for the product you been using for over a year, using the excuse it "might" one day in the future be damaged by a potential defect you have no evidence is affecting your device!

Okay I take it back. I will defend Intel a little here but quickly add, if the tables were turned and if these were AMD processors, I would pose the same defense.

There are ~ 3 billion (with a "b") transistors in the typical i7 processor. There are millions of lines of micro-code in the processor and associated chipset and BIOS firmware used to control that processor. I do not believe Intel (or AMD if the roles were reversed) is being purposely "deceptive" about being fully transparent about the problem. I believe that (1) they have to first and foremost "protect" the integrity of code to ensure security is not, in any way, compromised, (2) they have to protect any proprietary trade "secrets", and (3) due to the shear numbers of transistors and lines of code, they may not actually know the exact structure of the bug, or how to squash it without impacting some other feature, or without introducing some other bug or worse, introducing a security vulnerability.

You have to prove there's an issue to be able to RMA things.

You would think that would be obvious to everyone. I guess not.

 

[ShrimpBrime]

First thing I'd do is get a contact frame and maybe for ease of use get a kryosheet that fits your processor 38x38.

If that doesn't work

The second thing I'd do is from a seller with an excellent return policy buy a known good cooler just for testing.

If that doesn't work

The last thing I'd do also fron a seller with a good return policy is buy a 13700k and test to make sure both cpus behave the same.

Maybe none of those steps is an option for you but I'd at least get a contact frame.

Obviously after you've dialed in the bios which honestly a user shouldn't even have to do just to get their cpu to not hit 100c. That being said it really comes with the territory of owning a high end 13th/14th generation part they've always shipped with dubious out of box settings. Ryzen X parts aren't much different they all hit 95c and the X3D parts around 85c.... They just aren't degrading themselves at least as rapidly.

Also unless someone is using your exact cooling/case/motherboard/cpu/ambient comparing temps is mostly useless you could've just lost the lottery on both Cpu quality and the IHS ability to transfer heat if you purchased 10 13700ks they'd all run a bit differently throw in the fact that some degrade themselves and it's really a crap shoot trying to compare with anyone else's numbers.

That being said I wouldn't jump ship at least not right now I'd wait till AMD/Intel offers you a meaningful upgrade while ditching this troublesome platform assuming none of the above steps fixes the issue in a way that is satisfactory to you.

He just needs a decent water loop for a cpu capable of 253w load at an instant (density).

My 14700K on air does the exact same thing and I can get pretty low ambient temps under 70°F no issues. The cpu at defaults still spikes 100c fairly quickly.

I'm open table b3nching. I can tell you the VRM package runs warm, if not to say hot by the end of a CBR24 run to say the very least.

Even with my water loop, 13700K would run like 91c at 253w defaults.

It's not to say, to actually overclock a Raptor Lake processor, one would need outstanding cooling, full custom loop and say to pass 5.7/5.8ghz wi5h 13th gen, full on chilled water loop.

Dark Rock pro 4 is just not an adequate cooler. Even with a contact frame, which all the 1700 processors I've had, never once purchased the contact frame. I don't think it would make a big difference only because of transistor density. I mean aside from the board warping from high mounting pressure cause some coolers and users may be a little aggressive in this aspect. Perhaps, perhaps not.

But something must be changed to create a result. And I'd start with looking at case flow, ambient temps, and definitely a water loop. (Opinion)

 

[oxrufiioxo]

He just needs a decent water loop for a cpu capable of 253w load at an instant (density).

My 14700K on air does the exact same thing and I can get pretty low ambient temps under 70°F no issues. The cpu at defaults still spikes 100c fairly quickly.

I'm open table b3nching. I can tell you the VRM package runs warm, if not to say hot by the end of a CBR24 run to say the very least.

Even with my water loop, 13700K would run like 91c at 253w defaults.

It's not to say, to actually overclock a Raptor Lake processor, one would need outstanding cooling, full custom loop and say to pass 5.7/5.8ghz wi5h 13th gen, full on chilled water loop.

Dark Rock pro 4 is just not an adequate cooler. Even with a contact frame, which all the 1700 processors I've had, never once purchased the contact frame. I don't think it would make a big difference only because of transistor density. I mean aside from the board warping from high mounting pressure cause some coolers and users may be a little aggressive in this aspect. Perhaps, perhaps not.

But something must be changed to create a result. And I'd start with looking at case flow, ambient temps, and definitely a water loop. (Opinion)

It's been about a year since I've had some Hands on with a 13700k I remember it being in the low 90s out of the box at 253w with a contact frame with a 360 aio after an hour of CB looped.... So yeah it's just the reality for sure.

I was more talking about piece of mind than any of those options making a huge difference especially a know good cooler for a 13700k.

That's why I said this is just the reality of owning the 13700k/13900k and 14th gen equivalents at least with out of the box settings on most boards.

 

[ShrimpBrime]

It's been about a year since I've had some Hands on with a 13700k I remember it being in the low 90s out of the box at 253w with a contact frame with a 360 aio after an hour of CB looped.... So yeah it's just the reality for sure.

I was more talking about piece of mind than any of those options making a huge difference especially a know good cooler for a 13700k.

That's why I said this is just the reality of owning the 13700k/13900k and 14th gen equivalents at least with out of the box settings on most boards.

Love it. It's like back in the day when you got an Intel Extreme chip and it just boiled water lol.

My air cooler throttles 14700K to 4.8ghz. So setting the multiplier fixed x48 does ok without other intervention of power settings. He could just lower the frequency of the cpu, VID on auto should lower its self as well.

Or just lowering the LLC might work. Wonder if he ever tried it yet??

 

[oxrufiioxo]

Love it. It's like back in the day when you got an Intel Extreme chip and it just boiled water lol.

My air cooler throttles 14700K to 4.8ghz. So setting the multiplier fixed x48 does ok without other intervention of power settings. He could just lower the frequency of the cpu, VID on auto should lower its self as well.

Or just lowering the LLC might work. Wonder if he ever tried it yet??

A lot of intel processors just ran generally cool out of the box even some of the more notorious ones. I remember when people first switched to ryzen and how boosty it could be and got nervous with it at 50c at desktop because even moving the mouse makes it jump lol. Like I said just the reality of where tech seems to be going we've seen this forever in the laptop market chips hitting 90c+.

Still with all the drama going on with intel right now I don't blame the OP for being concerned.... Online isn't always the best reference for is my cpu ok or not because most have completely different setups. I think it's ok to take some extra steps just to make sure your cpu is performing as it should though.

Now if you just don't like the platform and that is why you want to make the switch, fine do it.... I didn't like Z390 and switched to X570 not for extra performance but just to get off the platform. I still kept my Z390 system for a time it just got relegated to secondary PC and then replaced.

Most people under volt these days because companies are pushing the silicon too hard they are all doing it. We don't OC we tweak the product to behave how we want it to now. It really is no different on either side only the 7800X3D I would be ok running out of the box these days but you give up a ton of MT performance for that privilege.

 

[ShrimpBrime]

A lot of intel processors just ran generally cool out of the box even some of the more notorious ones. I remember when people first switched to ryzen and how boosty it could be and got nervous with it at 50c at desktop because even moving the mouse makes it jump lol. Like I said just the reality of where tech seems to be going we've seen this forever in the laptop market chips hitting 90c+.

Still with all the drama going on with intel right now I don't blame the OP for being concerned.... Online isn't always the best reference for is my cpu ok or not because most have completely different setups. I think it's ok to take some extra steps just to make sure your cpu is performing as it should though.

Now if you just don't like the platform and that is why you want to make the switch, fine do it.... I didn't like Z390 and switched to X570 not for extra performance but just to get off the platform. I still kept my Z390 system for a time it just got relegated to secondary PC and then replaced.

Most people under volt these days because companies are pushing the silicon too hard they are all doing it. We don't OC we tweak the product to behave how we want it to now. It really is no different on either side only the 7800X3D I would be ok running out of the box these days but you give up a ton of MT performance for that privilege.

Nah, no reason to be concerned. We have threads about other websites posting news.

What we DONT have, is people making threads here about failed, failing or instability with their Intel processors.

But making a switch, I'd upgrade the 13 to 14th gen refresh, not go a different platform and less performance, only at the cost of wattage. But since it's sheer conversion of energy to heat, he'll gain 200mhz at the same thermal design point.

Old Intel chips you overclocked. Man-ually. Like a man.

New Intel chips are already maxed clocked. Pre clocked bitch chips. Need LN2 to have any fun.

That's the only difference. What was the old brags? My 8700K did 5.2ghz with de-lid.... All the way back to gen 1 haha.

What hasn't changed is the need for outstanding cooling to achieve these kinds of clocks. Think about Intels 900 series. 6ghz man. On water cooling.

Who would ever think to need anything more than a Hyper 212 for this??

 

[evernessince]

I did that, and the result was: Cinebench instant crash after 5 seconds lol

Whats odd too is that you have a great case and fan setup for airflow so your temps really should be better even without an under-volt.

Have you tried removing the side panel of your case and running Cinebench? Your temps shouldn't drop much with the side panel off but if they do that indicates inadequate airflow. We are essentially ensuring that are what appears to be a good cooling setup is actually performing like it should, which eliminates a potential variable.

Otherwise I believe Bobaganoosh's recommendation on checking the mounting is excellent as are his recommendation of a thermal pad. Thermal pads provide an extremely consistent installation albeit with a small hit to temps but it does almost guarantee that the TIM between the IHS and cooler is optimally installed.

Also, Intel hasn't fixed anything yet, so any replacement would be just as likely to have the same issues.

Degradation occurs over time, 2 weeks vs 1 year is a big difference. I digress though, OP is making an effort on fixing their issue and that should be the focus right now.

I have to wonder about the potential of bias here, noting the AMD in your system specs.

And you have Intel. Everyone has one of two vendors and therefore there is no unbiased opinions possible here (or in general, people are not unbiased regardless of what they claim).

This is just a pointless ad hominem attack, one of two in your comment, that adds nothing the conversation and further derails the thread.

 

[eidairaman1]

Just move on

 

[iameatingjam]

I mean I kinda regret going with intel, after rma'ing one cpu and now running my replacement well under its rated spec (14700kf). But I'm also not in a hurry throw it out while its still working..... erm, partially cause I'm broke hehehe.

I did some tweaking in the bios to get voltage down even on my b board. I did it to protect the cpu but it also reduced heat so maybe it will help you. And if offsets are causing crashes thats fine there's no offsets here. Okay so starting from the intel defaults, I turned CEP off, Then I reduced the ai AC and DC loadline from 110 to 70. My performance seemed to tank any lower than this. Idk if thats undervolt protection kicking in or what. Anyway you may be able to get even better results by reducing the AC loadline further and then increasing LLC but thats not what I did so I'll just tell you what I did.

So AC/DC loadline at 70 then, and this is probably the most important part, set an ai voltage cap, somewhere between 1300 and 1500, I put mine at 1300 (1.3v). This will stop the voltage spikes that are at the very least contributing to raptor lake's problems. I can only tell you how to get to it on Gigabyte boards unfortunately. Hopefully you have one. Its tweaker>advanced voltage settings>cpu/vrm settings>internal vr control>ia voltage limit. You will find those AC/DC LL values in that internal vr control section as well. If you don't have gigabyte.... I guess you'll either have to poke around, read the manual or google it perhaps. (EDIT: I see you have asus, sorry, I'm not sure how to do it on asus boards)

Okay next, I turned TVB off and underclocked my pcores to 50x and my ecores to 35x. I put pl1 and pl2 at 200w. I left llc on auto, that seems to work best for me.

Next I turned my memory speed down. High memory speed does stress the memory controller which is part of the cpu you know so its not entirely irrelevant even for your heat problem ( though probably not a large factor). Either way its a good idea to keep your cpu alive. You could turn xmp off entirely, but I decided to just reduce it to intel maximum supported value which is 5600mts.

All this combined, reduced my score in passmark cpu test from 55,000 to 50,000. So a 10% loss in performance. but it reduced my vcore under load from 1.48v to 1.2 with VID around the same-ish. And even under all core loads we are talking low 70s for temperature. I cannot tell a difference in gaming or general computing. But I game 4k so keep that in mind. 1080p might be a different story idk.

 

[oxrufiioxo]

I mean I kinda regret going with intel, after rma'ing one cpu and now running my replacement well under its rated spec (14700kf). But I'm also not in a hurry throw it out while its still working..... erm, partially cause I'm broke hehehe.

I did some tweaking in the bios to get voltage down even on my b board. I did it to protect the cpu but it also reduced heat so maybe it will help you. And if offsets are causing crashes thats fine there's no offsets here. Okay so starting from the intel defaults, I turned CEP off, Then I reduced the ai AC and DC loadline from 110 to 70. My performance seemed to tank any lower than this. Idk if thats undervolt protection kicking in or what. Anyway you may be able to get even better results by reducing the AC loadline further and then increasing LLC but thats not what I did so I'll just tell you what I did.

So AC/DC loadline at 70 then, and this is probably the most important part, set an ai voltage cap, somewhere between 1300 and 1500, I put mine at 1300 (1.3v). This will stop the voltage spikes that are at the very least contributing to raptor lake's problems. I can only tell you how to get to it on Gigabyte boards unfortunately. Hopefully you have one. Its tweaker>advanced voltage settings>cpu/vrm settings>internal vr control>ia voltage limit. You will find those AC/DC LL values in that internal vr control section as well. If you don't have gigabyte.... I guess you'll either have to poke around, read the manual or google it perhaps. (EDIT: I see you have asus, sorry, I'm not sure how to do it on asus boards)

Okay next, I turned TVB off and underclocked my pcores to 50x and my ecores to 35x. I put pl1 and pl2 at 200w. I left llc on auto, that seems to work best for me.

Next I turned my memory speed down. High memory speed does stress the memory controller which is part of the cpu you know so its not entirely irrelevant even for your heat problem ( though probably not a large factor). Either way its a good idea to keep your cpu alive. You could turn xmp off entirely, but I decided to just reduce it to intel maximum supported value which is 5600mts.

All this combined, reduced my score in passmark cpu test from 55,000 to 50,000. So a 10% loss in performance. but it reduced my vcore under load from 1.48v to 1.2 with VID around the same-ish. And even under all core loads we are talking low 70s for temperature. I cannot tell a difference in gaming or general computing. But I game 4k so keep that in mind. 1080p might be a different story idk.

When I was doing my own very unscientific testing gaming performance only dropped 5-8% in the 5 games I tested at 125w so not enough even with a 4090 to blindly tell it was different even at 1080p. That was a 14900k but it shouldn't be worlds different.

I think I would shoot for 5.5-5.8ghz though but that's me, otherwise if it is working great for you that is awesome and hopefully helpful to the OP.

10% loss is definitely worth the massive drop in temps and likely almost 0 loss in gaming at 4k though.

The good thing at least is both you an the OP got 2 years of extra warranty out of this whole ordeal and while it remains to be seen how intel handles RMA's I'd like to believe they will take care of their i7 and i9 owners who are likely to buy future products from them if they feel taken care of.

 

[Vario]

Could swap for 12900K instead? Lots of 12900KS on eBay for cheap. Theres a seller selling tray 12900KS for ~$300 and the 13700K sells for around $250-300, so it wouldn't cost you much more than the hassle (and fees).

Unlike the 7800X3D, its drop in to your current board.

 

[oxrufiioxo]

Could swap for 12900K instead? Lots of 12900KS on eBay for cheap. Theres a seller selling tray 12900KS for ~$300 and the 13700K sells for around $250-300, so it wouldn't cost you much more than the hassle (and fees).

Unlike the 7800X3D, its drop in to your current board.

I would def do that over swapping to a 7800X3D and also needing a new board not like decent Am5 boards are all that cheap.... But I would still ride it out with the 13700K for now....

 

[ShrimpBrime]

I mean I kinda regret going with intel, after rma'ing one cpu and now running my replacement well under its rated spec (14700kf). But I'm also not in a hurry throw it out while its still working..... erm, partially cause I'm broke hehehe.

I did some tweaking in the bios to get voltage down even on my b board. I did it to protect the cpu but it also reduced heat so maybe it will help you. And if offsets are causing crashes thats fine there's no offsets here. Okay so starting from the intel defaults, I turned CEP off, Then I reduced the ai AC and DC loadline from 110 to 70. My performance seemed to tank any lower than this. Idk if thats undervolt protection kicking in or what. Anyway you may be able to get even better results by reducing the AC loadline further and then increasing LLC but thats not what I did so I'll just tell you what I did.

So AC/DC loadline at 70 then, and this is probably the most important part, set an ai voltage cap, somewhere between 1300 and 1500, I put mine at 1300 (1.3v). This will stop the voltage spikes that are at the very least contributing to raptor lake's problems. I can only tell you how to get to it on Gigabyte boards unfortunately. Hopefully you have one. Its tweaker>advanced voltage settings>cpu/vrm settings>internal vr control>ia voltage limit. You will find those AC/DC LL values in that internal vr control section as well. If you don't have gigabyte.... I guess you'll either have to poke around, read the manual or google it perhaps. (EDIT: I see you have asus, sorry, I'm not sure how to do it on asus boards)

Okay next, I turned TVB off and underclocked my pcores to 50x and my ecores to 35x. I put pl1 and pl2 at 200w. I left llc on auto, that seems to work best for me.

Next I turned my memory speed down. High memory speed does stress the memory controller which is part of the cpu you know so its not entirely irrelevant even for your heat problem ( though probably not a large factor). Either way its a good idea to keep your cpu alive. You could turn xmp off entirely, but I decided to just reduce it to intel maximum supported value which is 5600mts.

All this combined, reduced my score in passmark cpu test from 55,000 to 50,000. So a 10% loss in performance. but it reduced my vcore under load from 1.48v to 1.2 with VID around the same-ish. And even under all core loads we are talking low 70s for temperature. I cannot tell a difference in gaming or general computing. But I game 4k so keep that in mind. 1080p might be a different story idk.

For you fellas and ladies out there with 13th and 14th Gen processors don't have to try hard to run the chips cool where you can actually manage the thermals with under-dog cooling.

So just to give you an example, I will give 1 screen shot. And give you the exact settings used. And this would be the starting point, not the finish point, I'll explain after setup details.

XMP enabled.
Cpu All cores P and E set manually to 40x multipliers fixed. Cache ratio 40x
EVERYTHING ELSE IS ON AUTO. Nothing else is changed in the bios.

As you will see here, Max core temp 79c and 70c package temp.
Increase multiplier P-Cores to 41x. You can increase E-cores too, but you go off testing from here.
No other interventions needed.

26k pts CBR23 ain't horrible. It's only 4ghz... Ambient temp 21c/70f. open table top.

Wraith Prism, no clamp.

 

[Outback Bronze]

So just to give you an example,

I ran my KS @ 5hgz on the P's and all 16 E's @ 3.2ghz just for testing purposes and I was lucky to hit 40°c. Pretty sure it was running under 1v. Couldn't believe how cool it ran. Given, I was on a 480 rad @ 10°c ambient

 

[AleXXX666]

Hello!

I have a Intel i7 13700k + Asus ROG STRIX Z790-H GAMING + Corsair Vengeance RGB DDR5 6000MHz 32GB, and after those problems with Intel degrading CPU's and now in the summer wasted of seeing that motherboard red light warning that my CPU is at 100º when i'm playing specific games are getting me wasted of having this combination and i refuse to buy an AIO just for this crappy hot CPU, when my Dark Rock Pro 4 is a beast of an air cooler.

My real question here is, what i do in my PC is 90% gaming, basically that, sometimes my Girlfriend uses my PC for some heavy renders at Archicad but it's very rare to happen, is it dumb to swap my current setup for a Ryzen 7 7800X3D and pair it with a motherboard like ATX Asus ROG Strix B650E-F Gaming WiFi, since it's new in features and already AM5 for future upgrades?

Thanks in advance for all opinions you can provide.

P.S. If this is the wrong section, please move it (System Builder's Advice??)

If you could afford, and spending that "delta" for change is not going to break your budget, I highly recommend. I'm also "pc hw curious" and have spent lots of $$$ to change my whole platforms just to try. Dumb & dumber, huh

I'm Intel-boy but currently I recommend highly AM5 (just that "normal" 7000 series of course without "hybrid tryings"), because Intel BS with e-cores which are useless in games is pure crap.

 

[AusWolf]

Since you already have a decent rig, waiting for the Intel microcode update might be worth it. With that said, I can't answer this question for you, I don't think anyone can.

One thing is for sure: the Dark Rock Pro 4 is more than enough to cool the 7800X3D. I have the non-Pro Dark Rock 4, and mine only goes up to 83-84 degrees in all-core loads.

 

[Taraquin]

I would undervolt and set a max powerusage