7800X3D will be an utter failure of a CPU

LittleBuddy said:
Memory tuning will be more relevant towards the end of DDR5's lifecycle when we have faster CPUs and GPUs. Right now, yea it's pointless.
Click to expand...
On 3D V-cache chips not so much. The rest of AMD's line-up somewhat, and Intel's line-up absolutely. Gains can be seen if you are willing to spend or become a pro RAM'er(R). Although I don't know at what point the bandwidth will top out for those chips and stifle. It could be now?
 
izzypizzy said:
Just for what it’s worth, since I shared one of GN’s tweets-

I’m not familiar with GN, never seen any of his content. I have been out of the game for ~10 years.

I don’t know anything about his content, and I don’t care to shit on AMD. I am new to AMD with my latest build and I’m totally stoked on the components that I have from them. But I am curious to hear about his findings because I feel like everything that’s been reported so far has been vague and coming from a veil of corporate protection.

My 7800x3d has been nothing short of fantastic. Really just curious is all.
Click to expand...

Welcome back... bear this in mind as you rejoin the community.

Things have changed and some of us are old(er) people now. We aren't busting companies like we did back in 1999 or 2000. We're tired. We're running stock. We don't give a flip.

Most of the written tech press, and real testing, has evaporated. This is because of "new media".

The new media "talking heads" generally suck. They are generally making content to get viewers rather than do something "substantive" in the industry.

That being said: At least Gamers Nexus TRIES to get it right. Sometimes they do. But it's generally a wasteland.

Friends don't let friends be fanbois. Carry a light sack.
 
The short version: AM5 is a total mess.

AMD did not communicate everything it should have to motherboard vendors, so there are some stupid settings and missing safety features on all boards (for example the thermal shutdown point is set too high on most mobos from MSI, Asrock, Gigabyte). And SOC is indeed set way too high on Asus boards in general (higher than software readings) so it has been the culprit of many deaths, but not all.

The good news is the CPUs are not bad. The motherboards are, but it can all be fixed through software with good communication between AMD and partners and settings that make sense (and actually work because all those mobos tend to have settings that don't get saved or don't reset properly etc).

No panic needed but it's still a very bad showing by AMD and partners. Especially with how expensive motherboards have become.
 
kalston said:
The short version: AM5 is a total mess.

AMD did not communicate everything it should have to motherboard vendors, so there are some stupid settings and missing safety features on all boards (for example the thermal shutdown point is set too high on most mobos from MSI, Asrock, Gigabyte). And SOC is indeed set way too high on Asus boards in general (higher than software readings) so it has been the culprit of many deaths, but not all.

The good news is the CPUs are not bad. The motherboards are, but it can all be fixed through software with good communication between AMD and partners and settings that make sense (and actually work because all those mobos tend to have settings that don't get saved or don't reset properly etc).

No panic needed but it's still a very bad showing by AMD and partners. Especially with how expensive motherboards have become.
Click to expand...
I might agree if it wasn't for the fact that almost all motherboard makers for years have been overclocking and overvolting everything at what amounts to stock settings. It just happens that Asus is one of the worst offenders.
 
Yea but for example thermal shutdown point is a basic setting for many years now, and I can't believe they would mess this up in 2023. All motherboard makers are at fault though it seems. And AMD's communication remains amazingly bad, and not only with us customers lol.
 
Last edited:
funkydmunky said:
On 3D V-cache chips not so much. The rest of AMD's line-up somewhat, and Intel's line-up absolutely. Gains can be seen if you are willing to spend or become a pro RAM'er(R). Although I don't know at what point the bandwidth will top out for those chips and stifle. It could be now?
Click to expand...
In my experience games still use System RAM on X3D chips and still benefit from tighter timings.
 
Concerning yes. Guy desperately needs edits and TLDRs in his video. Keeps repeating himself over and over.

40 mins wasted of my life.

Biggest concern for me is that I never turn off my PC. I was worried about 4090 for a while but then it was fine. A 7800X3D is sufficient safety risk to avoid for now. Maybe a new version might be better.
 
I've had my 7800x running in a test configuration for 15 hours idleing (it's doing a copy in the background of my games drive). The average soc v is 1.021V, BUT there is a max voltage reading of 2.04V!
1682881629711.png
 
Slade said:
I've had my 7800x running in a test configuration for 15 hours idleing (it's doing a copy in the background of my games drive). The average soc v is 1.021V, BUT there is a max voltage reading of 2.04V!View attachment 567440
Click to expand...

Bug with the reading. User on reddit experienced the same thing. It's basically reporting it wrong by a 2x factor.
 
Slade said:
I've had my 7800x running in a test configuration for 15 hours idleing (it's doing a copy in the background of my games drive). The average soc v is 1.021V, BUT there is a max voltage reading of 2.04V!
Click to expand...
What board do you have?
You can always adjust the SoC to fixed value at max 1.2v or 1.25v if you're using expo. Find the menu of Volt SoC or VDDIO or any other volt settings related to SoC and set it at fixed value, preferably under 1.3v.
____________________________________________________________________________________________

As per Steve video, I can agree with him wholeheartedly.
Like I said in another post, the mobo vendors always set their AUTO volt (whether it's vcore or SoC) a lot higher than the chip needs, even in AM4 boards so users shall be aware of this.
Me myself always set fixed vcore or SoC voltage on the bios, the first time I boot into the bios.

And ASUS really fucked up with their 400W loose limit protection??!!!
How can their engineers, their QA/QC team, missed that?
What AM5 cpus need to have 400W of limit for CPU and SoC?
 
mjoeTW said:
Click to expand...

So Steve hypes it up as being layers beyond the SOC voltage, video drops, and it's confirmed to be the SOC voltage. Nice clickbait.
KickAssCop said:
Concerning yes. Guy desperately needs edits and TLDRs in his video. Keeps repeating himself over and over.

40 mins wasted of my life.

Biggest concern for me is that I never turn off my PC. I was worried about 4090 for a while but then it was fine. A 7800X3D is sufficient safety risk to avoid for now. Maybe a new version might be better.
Click to expand...
I usually just jump to the conclusion part of the video and play it back at 1.5x speed. If something sounds interesting in the conclusion, then I'll go back and watch the whole thing. This video is classic Steve where it feels like he rambles without a script. I'm sure he has talking points to follow, but I think he would benefit by writing down complete thoughts instead of trying to talk around the talking points off the cuff.
 
Armenius said:
So Steve hypes it up as being layers beyond the SOC voltage, video drops, and it's confirmed to be the SOC voltage. Nice clickbait.

I usually just jump to the conclusion part of the video and play it back at 1.5x speed. If something sounds interesting in the conclusion, then I'll go back and watch the whole thing. This video is classic Steve where it feels like he rambles without a script. I'm sure he has talking points to follow, but I think he would benefit by writing down complete thoughts instead of trying to talk around the talking points off the cuff.
Click to expand...
Did we watch the same video? It was SOC voltage for the CPU burning itself up, and for the motherboards burning up it was improper safeguard settings from the motherboard BIOS.
 
I summarized parts of the video.

There are multiple issues

Over voltage
Over volt protection not kicking in
Bios bugs that show wrong reading (e.g., bios was showing 1.1 but actual voltage was 1.4+)
Bios bugs that don’t save voltage settings correctly
Some bleed in circuitry in CPU silicon that can lead to over voltage frying the CPU

Either way new bios will be out soon so we gonna be gud.
 
mullet said:
Like I said EAS (early adopter syndrome) nothing new.
Click to expand...
ALEAS "AMD Late Early Adopter Syndrom" ? The most issues Alder Lake had when it came out was the socket / CPU contact problem and the XMP compatibility issues. If you had a high quality cooler with a backplate, there was quickly (within a month) a workaround solution for it, though not as good as using one of those brackets third-parties started selling within a few months of release.

Those were the two main issues on a brand new Intel platform that simultaneously had a big process node change (14nm to 10nm) and architecture change (p and e cores). And although still not excusable by Intel, relative to this they weren't that serious nor did it take a long time for there to be solutions.

Yet here we are over seven months later from AM5 launch. The XMP compatibility issues Alder Lake was having are the same ones AM5 was and is still having with XMP/EXPO. The 7900/7950X3D core management is still bad over two months later (Alder Lake P/E was working decent right out of the gate). And there are a host of BIOS issues that are killing CPUs and reducing longevity of others.

If this is still considered early adoption at this point of AM5, seems there are big differences between the definition on AMD and Intel platforms. Though it seems if people want full stability on either Intel or AMD, it is best to be on a cycle of just building a new PC at the end of each platform, ignoring anything in-between. That still didn't work for AM4 though, where AMD never did fix the USB disconnect bug after several years of hardware and software chipset updates.
 
Things are screwy right now. On my B650E-F latest UEFI the SOC runs at 1.280-1.288 and when I do a negative offset of 0.005 it goes down to 1.228 and still runs fine. I'm leaving it there to see if there is stability because there may not be any need it to be so high.
 
Anything with AMD is early adoption. Heck they are still trying to figure out FSR and RT effects. Should be done by 2030. 👀

Jokes aside, I definitely never buy anything that is V1 from AMD. This time I was thinking X3D is like V2 but actually it is still very much V1 so better to play the waiting game.

Also am too busy to tinker with PC. I leave it on since I don’t want to waste 10 seconds boot up time. Imagine me trying to adjust voltages and getting worried if my house will be on fire next time I show up at home.
 
I'm still on an older bios and am waiting for the newest one to not be a beta. Even then I'll probably hold of to see what the consensus is. For now I just disabled expo and clocked the memory by hand to match the expo primary timings. Secondary timings are auto on the memory. The CPU has a slight negative offset voltage. SOC is manual at 1.2v. This has been running fine for a few days and Saturday I picked up a reference 7900xt by Power Color and no issues seen so far with about 3 hours of gaming and 12 or so uptime hours on the PC.
CPUZ and HWINFO shot:
1682968752386.png
 
Tomorrow is the last day for me to decide whether to keep or return my 7800X3D, and honestly I don't know.


If it's a simple AGESA and BiOS fix that would certainly be great. With such small amounts of overvoltage causing these extreme problems, you do have to wonder what the safe amounts really are and how they impact longevity and stability.

Is the hardware prone to degradation regardless of how low the SOC voltages are? Will my 6000 CL30 kit work in 6 months time?
 
therealjustin said:
Tomorrow is the last day for me to decide whether to keep or return my 7800X3D, and honestly I don't know.


If it's a simple AGESA and BiOS fix that would certainly be great. With such small amounts of overvoltage causing these extreme problems, you do have to wonder what the safe amounts really are and how they impact longevity and stability.

Is the hardware prone to degradation regardless of how low the SOC voltages are? Will my 6000 CL30 kit work in 6 months time?
Click to expand...
There shouldn't be any risk to the RAM. What will you do if you return it... go intel 13th gen instead?
 
therealjustin said:
Tomorrow is the last day for me to decide whether to keep or return my 7800X3D, and honestly I don't know.


If it's a simple AGESA and BiOS fix that would certainly be great. With such small amounts of overvoltage causing these extreme problems, you do have to wonder what the safe amounts really are and how they impact longevity and stability.

Is the hardware prone to degradation regardless of how low the SOC voltages are? Will my 6000 CL30 kit work in 6 months time?
Click to expand...
NightReaver is correct. Also, your RAM kit is not even the issue. The problem is how the mainboards are doling out power to the SOC and why its spiking beyond thresholds. The mainboard is the issue as far as anyone can tell and no just the X3D parts are at risk here. Could be any 7000 series CPU given the correct (or incorrect rather) conditions to see a surge in SOC voltage.

If you are that on the fence, dump the AMD hardware and pickup an Intel 13Something and you should be fine. You don't need the 13900 unless you're looking for productivity exclusively. At least the Intel chips this generation don't explode!

You call, personally, I am very happy at this point that I don't have to contend with this issue. So, happy my adoption of AMD was a dumpster fire and pushed me to Intel for the time being.
 
therealjustin said:
Tomorrow is the last day for me to decide whether to keep or return my 7800X3D, and honestly I don't know.


If it's a simple AGESA and BiOS fix that would certainly be great. With such small amounts of overvoltage causing these extreme problems, you do have to wonder what the safe amounts really are and how they impact longevity and stability.

Is the hardware prone to degradation regardless of how low the SOC voltages are? Will my 6000 CL30 kit work in 6 months time?
Click to expand...
Build it. Lock SOC @ <1.25V, don't use unlocked PBO til it's patched.
 
I Flashed my Asus to bios 1401 beta, loaded EXPOII Profile and lowered (manually, ymmv):

Vsoc = 1.20
VDDIO/MC = 1.288

I'm still verifying other voltages. Nothing else seems too out of wack yet. Yeah, PBO off in bios. Turned off all the auto OC shit.
 
I turned PBO disabled and I got worse package wattage. Turned PBO to eco-mode 65w and it went back down.

Mega6 try eco-mode 65 see what happens.
 
mullet said:
I turned PBO disabled and I got worse package wattage. Turned PBO to eco-mode 65w and it went back down.

Mega6 try eco-mode 65 see what happens.
Click to expand...
My PPT is 15% of 230 or 34.5w right now. I'm good on the power. Not turning in my "X".
 
You know more about it them me BUT I get 5.4 Ghz @ 90watts total draw @ 100% total load. I was just wondering what what your Ghz tops out at eco-mode 65 watts. Do you really give up your X are your benchmarks the same?

I can put my 7900 into X mode and the only thing I noticed is more watts and heat.
 
mullet said:
You know more about it them me BUT I get 5.4 Ghz @ 90watts total draw @ 100% total load. I was just wondering what what your Ghz tops out at eco-mode 65 watts. Do you really give up your X are your benchmarks the same?

I can put my 7900 into X mode and the only thing I noticed is more watts and heat.
Click to expand...
100% across the board is not one of my use cases. Dropping the voltages just enough so I can run my games 24/7/365 that's all.
 
You can power limit Intel to around 125w for gaming and still retain 98% of the performance (productivity suffers more if heavily multi-threaded). Will be very easy to keep cool and quiet at that wattage, even on air.
 
therealjustin said:
I guess the 13700K, but I hate the thought of that much heat and power usage.
Click to expand...
I will use 7800x3d, get B650 non-E chipset, and set the volt SoC / VDDIO to 1.2v max.
The cause of x3d chip to death is the SoC is set auto to 1.4v if the EXPO profile is enabled.

You can still enable EXPO, and manually set the volt SoC / VDDIO to 1.2v max.
 
riev90 said:
I will use 7800x3d, get B650 non-E chipset, and set the volt SoC / VDDIO to 1.2v max.
The cause of x3d chip to death is the SoC is set auto to 1.4v if the EXPO profile is enabled.

You can still enable EXPO, and manually set the volt SoC / VDDIO to 1.2v max.
Click to expand...
Does the system pass stability tests with this approach? Which tests have you performed to confirm this hypothesis?
Just asking as I am actively contemplating a 7800X3D and if this is that simple of a fix then I might as well do it sooner rather than later.
 
So far I've had my 7800x3d/asus 670e gaming build running on the beta asus bios 1401 with expo on and manual voltage settings for cpu since Saturday. The pc hasn't been doing much except getting rebuilt to be like my old system so when I swap the board into my current setup, it'll be just a few driver updates for the 4090 and minor configuration for my mouse and keyboard. Restoring all the gaming stores has been a mini nightmare. Epic store being the worst offender. Steam easiest being a copy/paste and it works job. A few games during copy had cannot read from source disk errors meaning some of those games had bad files and needed game verification.

I'm getting a proper AM5 cooler for it later this week which is another reason I'm doing the slow transition, but so far it's passed stability tests. I haven't really pushed the cpu yet as I've not set pbo, so essentially it's running stock with expo for memory at ddr 6000.
 
You must log in or register to reply here.
Back
Top