Intel processors crashing Unreal engine games (and others)

[DAPUNISHER]

No you do not get to choose. They will send you back the exact same model. You can sell it and put the cash toward 12th though. It should be a BNIB you get back.

 

[Turbonium]

No you do not get to choose. They will send you back the exact same model. You can sell it and put the cash toward 12th though. It should be a BNIB you get back.

Would you just "start fresh" with the non-damaged Raptor Lake? Or do what I'm planning?

I figure I'm doing the smart thing.

 

[DAPUNISHER]

Stable PCs don't crash no matter what you throw at them.

looks over at one of my dumbbells; igor are you certain about that?

Would you just "start fresh" with the non-damaged Raptor Lake? Or do what I'm planning?

I figure I'm doing the smart thing.

I'd swap for 12th and sell the raptor simply because we have no idea what the true extent of the issues are. You need this for a workflow, and all of the data I have seen shared says Alder Lake is rock solid with standard industry failure rates.

 

[igor_kavinski]

I'd swap for 12th and sell the raptor simply because we have no idea what the true extent of the issues are. You need this for a workflow, and all of the data I have seen shared says Alder Lake is rock solid with standard industry failure rates.

Seconded. God knows I tried to kill my 12700K (as an excuse to buy something new), even with the AC off and sweating a bit and all it did was shut down. I was OFFENDED!

 

[jpiniero]

Buildzoid shows the transient 1.6v spikes and goes over gigabyte bios settings to limit VID requests from the intel cpu.

setting change at the 12min mark.
limiting to 1.4v disabled the 6ghz boost on 2 cores, loses 1000 points on cinebench23.

That's kind of what I was wondering about.

Feels like it's too easy to just cap the voltage.

 

[lakedude]

Would you just "start fresh" with the non-damaged Raptor Lake? Or do what I'm planning?

I haven't kept up with the improvements since 12th gen but there is a good chance that 12th gen is missing something you paid for and might actually want. Surely they made something better in 2 generations?

 

[DAPUNISHER]

I haven't kept up with the improvements since 12th gen but there is a good chance that 12th gen is missing something you paid for and might actually want. Surely they made something better in 2 generations?

Bruh, we are in a thread about raptor CPUs suffering degradation and becoming unstable. There is nothing better about that, nothing.

EDIT: they didn't change much, other than fix what ain't broke, and crank it up to 11.

 

[igor_kavinski]

Surely they made something better in 2 generations?

Increased cache amounts a bit, added more E-cores and upped the frequencies. And "refined" the Intel 7 process technology to hit higher speeds which as we can now see, wasn't really a refinement.

 

[ondma]

There is at least one major S.I. that is failing 10-25% of raptor during their testing now. Prebuilts under warranty can and do have the whole system shipped back.

Tech tubers doing the secret shopper, have been railing about Dell charging them for the premium warranty even after explicitly rejecting the offer during ordering. I hope that slimy practice costs them big over this. If the system is out of warranty that company's reputation is the one that will suffer most, not Intel's. It'll catch up to them financially in lost orders and sabotaged relationships, if not credits for incurred expenses. Or maybe all of them. If your Dell/Alienware is nerfed a couple years after bought it, and they want to charge you $$$ to repair it, few will be ordering their next PC from them.

Users savvy enough to ask for help online will discover they have a defective Intel CPU.

You call it hate. I call it accurate observations describing decades of anti competitive and anti consumer practices. This is but the latest example. I'd say SSDD, but things are different this time, all the usual tactics of whataboutism, astroturfing, gaslighting, and shooting the messenger are failing miserably. Nvidia won a whole generation of gamers by executing at the right time while AMD kept stumbling along and failing to keep up. From where I am sitting, that dynamic is playing out in CPUs now. Intel is cementing a bad rep with the zoomers that will haunt them for years and years.

Who's going to go with Intel for budget when the upgrade path is so limited, and every tech tuber worth their salt is telling them not to buy Intel? No one is going to have any confidence in Bartlet if it does come to desktop, and it is starting to look like even money or better they won't even try to make that happen.

I don't want to derail the thread, but the name calling and personal insults directed to intel users is inexcusable. I wont infract myself by using names, but it certainly does happen by posters who should know better.

 

[igor_kavinski]

I don't want to derail the thread, but the name calling and personal insults directed to intel users is inexcusable.

Click "Report".

 

[lakedude]

Bruh, we are in a thread about raptor CPUs suffering degradation and becoming unstable. There is nothing better about that, nothing.

Yes, they are obviously not better in their current state.

EDIT: they didn't change much, other than fix what ain't broke, and crank it up to 11.

Really? Same socket and same RAM? No faster PCI bus or next gen USB? No AVX improvements? No security improvements? Nothing? (I'm actually asking as it would be a surprise to me)

 

[igor_kavinski]

Really? Same socket and same RAM? No faster PCI bus or next gen USB? No AVX improvements? No security improvements? Nothing? (I'm actually asking as it would be a surprise to me)

Raptor's replacement was Meteor Lake S which fell short of their expectations and Arrow Lake was too far away so they had nowhere to turn to but rehash Raptor Lake with a Refresh.

 

[ondma]

Yes, they are obviously not better in their current state.


Really? Same socket and same RAM? No faster PCI bus or next gen USB? No AVX improvements? No security improvements? Nothing? (I'm actually asking as it would be a surprise to me)

Raptor did include some increased/improved cache, I believe, that increased IPC slightly and improved gaming, but Intel simply pushed the envelope to far in going for max clock speeds, although since non-K models now seem affected, there may be a deeper problem as well.

 

[igor_kavinski]

although since non-K models now seem affected, there may be a deeper problem as well.

Could be sabotage from a disgruntled employee (someone handling microcode, either the development or the validation of it) or engineers getting back at management for missed bonuses/raises etc.

 

[KompuKare]

Yes, they are obviously not better in their current state.


Really? Same socket and same RAM? No faster PCI bus or next gen USB? No AVX improvements? No security improvements? Nothing? (I'm actually asking as it would be a surprise to me)

The changes between Alder Lake and Raptor Lake were so small (think the various Skylake revision/rebrands a few years ago), that the Wikipedia combined the P cores article (the E cores 100% the same but there was more of them on many SKUs):

Golden Cove - Wikipedia

en.wikipedia.org

WikiChip is a bit more generous:

Raptor Lake - Microarchitectures - Intel - WikiChip

Raptor Lake (RPL) is Intel's successor to Alder Lake, an enhanced Intel 7-process based microarchitecture for mainstream workstations, desktops, and mobile devices. The microarchitecture was developed by Intel's R&D center (IDC) in Haifa, Israel.

en.wikichip.org

My biggest concern with Intel for the last few gens has been the crazy Netburst-like power wastage.

GHz races, 300W at stock on some loads, etc. these are all things which the current mess will hopefully bring to an end. Remember, if one vendor "wins" a benchmark by 1-2% but uses twice the power it also encourages the other vendor to go reckless with power just to catch up. A vicious cycle which I hope we will now be spared.

But despite all this, far too many buyers were prepared to reward Intel for "winning" by a few %. The nobody ever got fired for Intel line is one thing, but I am sorry if unknowledgable buyers reward Intel for such poor products I think we are allowed to call them out on this.

That Intel have also been doing anti-competitive and anti-consumer things years make it hard to feel sorry for them.

 

[DAPUNISHER]

Click "Report".

This 100%. If it is directed at any member/s here it will be acted upon. We also don't allow any version of fanboy, stan. or the like. You can however refer to yourself in that manner, there is nothing wrong with liking what you like and owning it.


The first rule of fight club/tech forums is attack the post, not the poster.

 

[DAPUNISHER]

Scooby clue from a reddit post a couple of years ago?

https://www.reddit.com/r/intel/comments/10dcjd9/intel_blocks_undervolting_on_alder_and_raptor_lake

 

[ondma]

The changes between Alder Lake and Raptor Lake were so small (think the various Skylake revision/rebrands a few years ago), that the Wikipedia combined the P cores article (the E cores 100% the same but there was more of them on many SKUs):

Golden Cove - Wikipedia

en.wikipedia.org

WikiChip is a bit more generous:

Raptor Lake - Microarchitectures - Intel - WikiChip

Raptor Lake (RPL) is Intel's successor to Alder Lake, an enhanced Intel 7-process based microarchitecture for mainstream workstations, desktops, and mobile devices. The microarchitecture was developed by Intel's R&D center (IDC) in Haifa, Israel.

en.wikichip.org

My biggest concern with Intel for the last few gens has been the crazy Netburst-like power wastage.

GHz races, 300W at stock on some loads, etc. these are all things which the current mess will hopefully bring to an end. Remember, if one vendor "wins" a benchmark by 1-2% but uses twice the power it also encourages the other vendor to go reckless with power just to catch up. A vicious cycle which I hope we will now be spared.

But despite all this, far too many buyers were prepared to reward Intel for "winning" by a few %. The nobody ever got fired for Intel line is one thing, but I am sorry if unknowledgable buyers reward Intel for such poor products I think we are allowed to call them out on this.

That Intel have also been doing anti-competitive and anti-consumer things years make it hard to feel sorry for them.

Cache increased on ADL - RL from 14+30mb to 32+36mb (L2+L3), and the E cores increased on most models. Until the stability issues showed up, I considered it decent step up, sort of intermediate between a refresh and a new generation. I would consider it more significant than a Skylake type refresh.

RL-RLR just pushed the clock speeds even further, like the Skylake refreshes. It became pretty obvious that Intel was not executing well on the desktop when RL-R was released. I had hoped ARL would be a big jump in performance and power efficiency to make up for slipping back after ADL, but it does not appear to be.

 

[KompuKare]

Scooby clue from a reddit post a couple of years ago?

https://www.reddit.com/r/intel/comments/10dcjd9/intel_blocks_undervolting_on_alder_and_raptor_lake

Well there goes this poster on another forum who was asking about advice to tame their 14900HX laptop - certainly looks like even software undervolting would be not be possible.

Now Intel must have had their reasons. Undervolting can cause instability but the few people tinkering with undervolting should be aware of that and would just dial back so Intel's reputation would not suffer.

So, more likely is that a few years ago they noticed that their boost algorithm is flawed and

1. raises clocks too high when undervolted,
2. raises voltage (a bit ironic),
3. or the way it takes account of voltage, temp, and clocks is flawed

Or some completely other reasons... but does any one believe in coincidences?

Cache increased on ADL - RL from 14+30mb to 32+36mb (L2+L3), and the E cores increased on most models. Until the stability issues showed up, I considered it decent step up, sort of intermediate between a refresh and a new generation.

RL-RLR just pushed the clockspeeds even further.

I'd put that down as a refresh. In the days of 3D stacked cache from the other vendor, increasing L3 cache is even a packaging thing. Not touching the cores ... tinkering around the cores ... is what during the Skylake stagnation.

 

[In2Photos]

I have a Dell XPS 15 laptop with an Intel 9750H. I can't stand the sound of the fans running. So I looked into undervolting. Intel actually applied a BIOS update that blocked it. Thankfully I was able to revert the BIOS update and use ThrottleStop to both undervolt and limit boosting to keep this thing a little quieter. The fact that they blocked it on flagship CPUs is mind boggling!

 

[coercitiv]

Now Intel must have had their reasons.

Security, there was an exploit using undervolting.

Two teams of researchers—one at the University of Birmingham in the UK, TU Graz in Vienna, KU Leuven in Belgium and another at the Technische Universität Darmstadt in Germany and the University of California—have found a new technique that can allow hackers to fiddle with the voltage of Intel chips to cause them to leak information stored using Intel's Secure Guard Extensions feature. Those "secure enclaves" in a device's memory are designed to be impregnable. Intel, which asked the teams to keep their findings under wraps for the last six months, confirmed the findings and pushed out an update to its chip firmware to prevent the attack today.

The technique, which one of the two teams calls Plundervolt, involves planting malicious software on a target computer that temporarily reduces the voltage of the electricity flowing to an Intel chip. That drop in voltage, known as "undervolting," typically allows legitimate users to save power when they don't need maximum performance. (By that same token, you can use the voltage-variance feature to "overclock" a processor for more intensive tasks.) But by momentarily undervolting a processor by 25 or 30 percent, and precisely timing that voltage change, an attacker can cause the chip to make errors in the midst of computations that use secret data. And those errors can reveal information as sensitive as a cryptographic key or biometric data stored in the SGX enclave.

 

[KompuKare]

Security, there was an exploit using undervolting.

Thanks, still seems like a sledgehammer for a nut response. Users would have far preferred them making any request to undervolt from software far slower and 100% superuser. But then security features have been robbing users of performance and features a lot in recent years.

 

[DAPUNISHER]

Could be sabotage from a disgruntled employee (someone handling microcode, either the development or the validation of it) or engineers getting back at management for missed bonuses/raises etc.

This isn't going to help - Intel to Cut Thousands of Jobs to Reduce Costs, Fund Rebound

E.U. members on PCMR confirming they should have recourse with the retailers. And that most or all of those businesses have to take it up with Intel to get compensated, because regardless, they will have to replace or refund. I am certain the retailers are all delighted about that prospect.

Most of the top posts there when sorting by hot are about raptor at the moment. With 10s of thousands of upvotes between them; Streisand effect indeed.

 

[Saylick]

This isn't going to help - Intel to Cut Thousands of Jobs to Reduce Costs, Fund Rebound

Yikes. I guess those jobs are in the rearview mirror, too.

 

[positivedoppler]

This isn't going to help - Intel to Cut Thousands of Jobs to Reduce Costs, Fund Rebound

E.U. members on PCMR confirming they should have recourse with the retailers. And that most or all of those businesses have to take it up with Intel to get compensated, because regardless, they will have to replace or refund. I am certain the retailers are all delighted about that prospect.

Most of the top posts there when sorting by hot are about raptor at the moment. With 10s of thousands of upvotes between them; Streisand effect indeed.

I remember AMD used to have mass layoffs right before a disastrous earning report. Maybe Intel is stalling to address the mass recall or refund decision until after their Thursday earning report because if earning is bad and then you pile on most of the CPU sold in the past 1.5 years will have to be recalled...omg