|
I just built a new PC with an i5-13600K, ASUS Strix Z790-A II, and TeamGroup T-Create 2x16GB 6000 CL30, and I'm getting memory errors with XMP after 2 hours of TestMem5 with Anta Absolut config. The RAM SPD Hub temp gets up to 55C when it errors. If I let it reach 58C it errors in under 10 minutes. Is that too hot for DDR5? Could it be bad memory or CPU IMC? My motherboard auto sets the VCCSA to 1.248V and the IMC to 1.332v, is that normal for XMP? I expected DDR5 6000 XMP to be an error-free experience on this system and I'm not looking forward to hours or days of troubleshooting. Should I just buy a 5600 JEDEC kit and not worry about it?
|
# ? Dec 10, 2023 07:10 |
|
|
# ? Jun 13, 2024 05:06 |
|
LegitMaan posted:I just built a new PC with an i5-13600K, ASUS Strix Z790-A II, and TeamGroup T-Create 2x16GB 6000 CL30, and I'm getting memory errors with XMP after 2 hours of TestMem5 with Anta Absolut config. Your voltages are fine but those temps are quite high. Do you have any airflow over the ram at all? Typically, instability at higher temperatures is due to agressive tRFC/tREFI timings which some motherboards have been known to push when you enable XMP. The JEDEC timings are rated up to a memory die temp (which is typically higher than the SPD temp) of 85C, but if you have any airflow at all it's possible to use more aggressive values for a pretty noticeable performance gain and manufacturers like to exploit this. Could you grab the Asus tools pack here. This link is sourced from the lead BIOS dev here. There should be a tool called "Memtweakit" that will let you see all the memory timings in a convenient fashion from windows. If you post a screenshot here I'll be able to tell you if there is anything egregious. If you want to manually set these timings to the JEDEC values in the BIOS, tRFC should be 350ns (which is 1050 cycles at 3000MHz) and tREFI should be 7.8us (23400 cycles at 3000MHz).
|
# ? Dec 10, 2023 07:55 |
|
BurritoJustice posted:Your voltages are fine but those temps are quite high. Do you have any airflow over the ram at all? I have my system in a Fractal North with the front fans running at around 900 RPM and in a room that's about 77 F. I didn't think that would lead to overly high memory temperatures. I attached the memtweakit stats.
|
# ? Dec 10, 2023 08:09 |
|
LegitMaan posted:I have my system in a Fractal North with the front fans running at around 900 RPM and in a room that's about 77 F. I didn't think that would lead to overly high memory temperatures. Your motherboard has automatically tuned tRFC down to 160ns, which Hynix DDR5 should be perfectly able to do but it definitely can cause issues once your temperatures go above 50C. In the BIOS, I'd change that to 600 cycles (200ns) and retest for stability. I dunno why they've tuned tRFC so tight considering the other subtimings are not tight at all. In terms of the high temps, could be due to the North having no top fans next to the ram combined with using an AIO cooler (just a guess to what you're using, as tower coolers provide memory airflow). Or maybe the TeamGroup heatspreaders are just awful for some reason.
|
# ? Dec 10, 2023 08:26 |
|
BurritoJustice posted:Your motherboard has automatically tuned tRFC down to 160ns, which Hynix DDR5 should be perfectly able to do but it definitely can cause issues once your temperatures go above 50C. In the BIOS, I'd change that to 600 cycles (200ns) and retest for stability. I dunno why they've tuned tRFC so tight considering the other subtimings are not tight at all. Thanks for the info. I'll try adjusting the tRFC tomorrow and do some more tests. Oh I forgot to mention the cooler, its a Noctua D15S and the fan is going at around 800 RPM.
|
# ? Dec 10, 2023 08:48 |
|
LegitMaan posted:Thanks for the info. I'll try adjusting the tRFC tomorrow and do some more tests. Oh I forgot to mention the cooler, its a Noctua D15S and the fan is going at around 800 RPM. Yeah I have no clue why your memory is running so hot with a fairly vanilla 1.35V XMP profile, I'll be honest.
|
# ? Dec 10, 2023 08:52 |
|
BurritoJustice posted:Yeah I have no clue why your memory is running so hot with a fairly vanilla 1.35V XMP profile, I'll be honest. I have no idea either. It idles in the low 30s, but TM5 Absolut pushes it up to the mid-high 50s where it errors. Strangely, I ran Karhu RAM test and it passed 30,000% with the RAM hitting 58C no problem. Is DDR5 just that finicky and temperature sensitive? Edit: Also, how likely is it that my problem is the CPU's IMC or the motherboard? I'm willing to swap the RAM but I don't want to go on a wild goose chase with the other stuff. Edit 2: Setting tRFC to 600 didn't help, still got an error at around 59C. Curiously I set my memory to auto in the BIOS to get the stock 4800 40-40-40-76 specs and the tRFC is only 383, is that normal? LegitMaan fucked around with this message at 10:02 on Dec 10, 2023 |
# ? Dec 10, 2023 09:10 |
|
LegitMaan posted:Edit: Also, how likely is it that my problem is the CPU's IMC or the motherboard? I'm willing to swap the RAM but I don't want to go on a wild goose chase with the other stuff. Very unlikely, DDR5-6000 is low enough for Intel that every CPU and motherboard should be able to handle it fine. LegitMaan posted:Edit 2: Setting tRFC to 600 didn't help, still got an error at around 59C. Curiously I set my memory to auto in the BIOS to get the stock 4800 40-40-40-76 specs and the tRFC is only 383, is that normal? No, JEDEC 4800 should be 840 for tRFC. I'm guessing it won't get hot enough at stock 1.1v for it to be an issue though. If you want to make sure that it is due to high temps causing refresh problems, you can enable XMP then set the tRFC to the JEDEC value of 350ns (which should be 1050 cycles in the bios). Asus seems to be gassing the tRFC at stock to win motherboard benchmarks at the cost of high temp stability.
|
# ? Dec 10, 2023 12:57 |
|
First die pictures of Meteor Lake showing the tile layout are coming out from an upcoming Samsung laptop, but the reason I am posting this is Samsungs thermal paste/putty application is driving people insane online lol:
|
# ? Dec 10, 2023 21:38 |
|
lol it’s like someone stuccoing their house and failing.
|
# ? Dec 10, 2023 21:39 |
|
risky frosting
|
# ? Dec 10, 2023 22:07 |
|
The “a serving size of cookies is a whole sleeve” people have discovered thermal paste.
|
# ? Dec 10, 2023 22:11 |
|
_ a dab in the middle _ a centered "x" _ a 3 x 3 grid of dots X however much cream cheese you think is sensible for a bagel, just go with your heart on this one
|
# ? Dec 10, 2023 22:31 |
|
Cygni posted:First die pictures of Meteor Lake showing the tile layout are coming out from an upcoming Samsung laptop, but the reason I am posting this is Samsungs thermal paste/putty application is driving people insane online lol: Which manufacturer had to recall a specific batch of laptops because some random employee was applying ten or twenty times as much thermal paste as the system actually called for?
|
# ? Dec 10, 2023 22:35 |
|
Potato Salad posted:Which manufacturer had to recall a specific batch of laptops because some random employee was applying ten or twenty times as much thermal paste as the system actually called for? Asus and then they just did the normal Asus support thing of making poo poo up to deny claims. Eg https://www.reddit.com/r/ASUS/s/qYLwM5t5IW
|
# ? Dec 10, 2023 22:45 |
|
Oh poo poo, that looks way more recent. And bad, because that's actually damage as opposed to just very poor thermals.
|
# ? Dec 10, 2023 23:00 |
|
I don’t think my first attempt to lay bricks looked that bad, and I suck as a bricklayer.
|
# ? Dec 10, 2023 23:41 |
|
Cygni posted:First die pictures of Meteor Lake showing the tile layout are coming out from an upcoming Samsung laptop, but the reason I am posting this is Samsungs thermal paste/putty application is driving people insane online lol: You can see marks on the top of the die. it looks like somebody used something to push all the paste to the sides, which means the actual application is probably not this bad.
|
# ? Dec 11, 2023 03:56 |
|
When I was recently looking at high-end gaming laptops (before ultimately settling for an $800 BF Lenovo), I noticed more than a few advertised liquid metal TIM as a selling point, and all I could think of is "yeah, I'm sure that's being applied with the utmost of care on a 'speed it up or get fired' assembly line in China."
|
# ? Dec 11, 2023 04:07 |
|
EoRaptor posted:You can see marks on the top of the die. it looks like somebody used something to push all the paste to the sides, which means the actual application is probably not this bad. That was my thought as well, it's not ideal but it's not catastrophically bad either.
|
# ? Dec 11, 2023 04:43 |
|
hobbesmaster posted:Asus and then they just did the normal Asus support thing of making poo poo up to deny claims. Eg is that better or worse than (MSI quality) FuturePastNow fucked around with this message at 05:52 on Dec 11, 2023 |
# ? Dec 11, 2023 05:48 |
|
FuturePastNow posted:is that better or worse than I've never understood why the big PC manufacturers are so awful at build quality, given that phones are for the most part better built, and the cheaper consoles have some pretty drat impressive build quality and design these days. https://www.ifixit.com/Teardown/Xbox+Series+X+Teardown/138451 I'd love to do a PC with an Xbox style cooling setup, vapor chamber and a 130mm fan that also pulls air across all the other components. It's roughly a shrunk down Torrent case.
|
# ? Dec 11, 2023 06:25 |
|
Twerk from Home posted:I've never understood why the big PC manufacturers are so awful at build quality, given that phones are for the most part better built, and the cheaper consoles have some pretty drat impressive build quality and design these days. Profit margins are bad in the PC world except for some super expensive stuff.
|
# ? Dec 11, 2023 06:37 |
|
The xbox designs are super cool, the series S is really nice too. There was some funky vertical case from corsair for the "Corsair One" but I think that was only for prebuilts and they didn't sell the case separately. It was still not anywhere near as nice as the series X for airflow from the exploded views I've seen though.
|
# ? Dec 11, 2023 06:43 |
|
Twerk from Home posted:I've never understood why the big PC manufacturers are so awful at build quality, given that phones are for the most part better built, and the cheaper consoles have some pretty drat impressive build quality and design these days. A specific Xbox model might sell more than 2 million units, all exactly identical. It makes sense to invest in the precision tooling, jigs, and automation for the production run, because reducing labour and return percentage can lead to big savings. For a large oem, like Dell or HP, they might expect to sell 500k ( or more ) of a popular laptop or desktop. They don’t build them, and outsource both the design and build to integrators based on a spec, so things can vary drastically. They absolutely can get very high quality, but they won’t get to the very top as investment isn’t worth it for a product that might only be on shelves for 6 to 8 months. There is also often design assistance/guidance from intel or AMD For a smaller oem, like Asus or MSI, one of their laptops might sell 100k to 200k, and investing in any custom design is too costly. Instead, they buy off the shelf designs from the same integrators that build for the big oems, and basically get spare time and space in older facilities. Every group ends up with a different path to optimizing costs and getting the most profit, and the smaller you are, the more it’s about cutting quality.
|
# ? Dec 11, 2023 07:33 |
|
There was a generation of (Intel) macbook with overheating issues due to too much paste and someone found that the service manual had the tech apply an entire tube of paste.
|
# ? Dec 11, 2023 14:56 |
|
in a well actually posted:There was a generation of (Intel) macbook with overheating issues due to too much paste and someone found that the service manual had the tech apply an entire tube of paste. Looks like it actually suggests one complete syringe per pad.
|
# ? Dec 11, 2023 15:37 |
|
Twerk from Home posted:
It should probably be one syringe per 3 pads. That’d make more sense.
|
# ? Dec 11, 2023 15:54 |
|
Twerk from Home posted:I've never understood why the big PC manufacturers are so awful at build quality, given that phones are for the most part better built, and the cheaper consoles have some pretty drat impressive build quality and design these days. When you make 100 different skus of crappy laptops there's no time to design, they are merely assembled out of the cheapest parts on hand that day.
|
# ? Dec 11, 2023 16:12 |
|
Twerk from Home posted:
Thank you, could not find that.
|
# ? Dec 11, 2023 17:28 |
|
just remembered the RX 7600 has half the cache of the other 7000 series on top of being an older node lol. It is funny that people think the alder lake chips having less cache is a big deal because literally every time I think about it there’s more examples - AMD’s APUs have less cache (and AMD uses APUs as the low-end desktop lineup), arm gives each architecture an explicit range that is customized for each product, etc. like it’s totally obvious that the people who are super upset about 14th gen have never looked at a spec sheet for a lineup before, the lower-end chips are gonna perform lower and cache is one of the obvious knobs to turn. Alder lake is basically the same as raptor other than the cache, so, who cares? 14th gen is bad because it’s not a performance step, and because of what a rebrand-gen says about intel’s health as a company, not because it’s deceptive, anymore than RX500 series was deceptive, or the GTX 700 series was deceptive, or APUs are deceptive etc. It seems like people collectively forgot that yearly cadence (with rebrands if needed) is a thing oems want and that and low-end SKUs exist. Same energy as the people who got big mad about smaller memory buses on Ada without realizing that AMD had done the same change the previous gen. Paul MaudDib fucked around with this message at 20:32 on Dec 11, 2023 |
# ? Dec 11, 2023 20:13 |
|
Paul MaudDib posted:14th gen is bad because it’s not a performance step, and because of what a rebrand-gen says about intel’s health as a company, not because it’s deceptive, anymore than RX500 series was deceptive, or the GTX 700 series was deceptive, or APUs are deceptive etc. It seems like people collectively forgot that yearly cadence (with rebrands if needed) is a thing oems want and that and low-end SKUs exist. Did anyone say that 14th was deceptive? I think everyone just said it was no more or less deceptive than the Ryzen they were bitching about. Now if you want actually deceptive, go back to that presentation and check the two CPUs that Intel is putting up bar charts for performance comparisons, and then go check the prices that laptops containing those CPUs are selling for. The intel one rarely sells for under $600, the AMD one rarely goes above $400. I would be there's a substantial different in tray cost for those 2, and if Intel had picked 2 equal-cost CPUs to compare the whole "latest" thing would be a moot point.
|
# ? Dec 11, 2023 22:07 |
|
I think you will find that latest doesn’t always mean best, unless it’s an i5 from intel, which means best because it’s the latest It’s that simple
|
# ? Dec 11, 2023 23:58 |
|
Worf posted:I think you will find that latest doesn’t always mean best, unless it’s an i5 from intel, which means best because it’s the latest I still regularly run across people who think that i3 and i5 are old and i7 and i9 are new, and will assume that an ancient i7 is faster than a current i5.
|
# ? Dec 12, 2023 00:10 |
|
Twerk from Home posted:I still regularly run across people who think that i3 and i5 are old and i7 and i9 are new, and will assume that an ancient i7 is faster than a current i5. Khorne fucked around with this message at 02:01 on Dec 12, 2023 |
# ? Dec 12, 2023 01:56 |
|
Is Meteor Lake mobile only? Just wondering why they'd reheat the 13th gen when they're about to release chips with actual advances.
|
# ? Dec 12, 2023 21:01 |
|
ConanTheLibrarian posted:Is Meteor Lake mobile only? Just wondering why they'd reheat the 13th gen when they're about to release chips with actual advances. yup, Meteor Lake will be mobile only and have its own new naming scheme. "14th gen" will only be Raptor Lake Refresh, and is the end of the road for that naming scheme on desktop as well. next desktop architecture is Arrow Lake which is expected end of next year e: looking further forward for desktop, after Arrow Lake is going to be an Arrow Lake Refresh in 2025 that is mostly about adding more ecores. there are a few more mobile specific designs on the horizon as well. there will be Lunar Lake (a small, high efficiency design) end of 24/start of 25, and Panther Lake (a bigger mobile design) end of 25/start of 26. 2026 is supposedly the first parts with "Rentable Units" Cygni fucked around with this message at 22:53 on Dec 12, 2023 |
# ? Dec 12, 2023 22:44 |
|
BurritoJustice posted:Very unlikely, DDR5-6000 is low enough for Intel that every CPU and motherboard should be able to handle it fine. TRFC at 1050 didn’t make a difference for stability so I set it back to 480 and raised memory voltage from 1.35V to 1.4V, which seemed to do the trick. TM5 passed 25 cycles despite the memory hitting 63C. Meanwhile at 1.35V it fails after 3 cycles at 54C. The error messages point to voltage and TRFC. It’s weird how it’s so temperature sensitive until the voltage is increased beyond spec. I’m thinking this kit is faulty at its rated voltage. Also, is there a resource for info on default TRFC and other sub timings for JEDEC speeds?
|
# ? Dec 13, 2023 18:21 |
|
Via is back, baby. I'm really surprised how dramatically they seem to have caught up, too: https://www.anandtech.com/show/21189/zhaoxin-unveils-kx7000-cpus-eight-x86-cores-at-up-to-370-ghz quote:Based on the company's Century Avenue microarchitecture, the processor features up to eight general-purpose x86 cores running at 3.70 GHz, while utilizing a chiplet design under the hood. Both DDR4 and DDR5 memory controllers, DX12 iGPU, and they're LGA 1700! Are these really chiplet processors built at TSMC or is there a chance that it's an Intel processor with the silk screen filed off like some other Chinese companies have done to meet domestic sourcing requirements? If you're not familiar, the fastest Chinese domestic branded client CPU was the PowerStar P3-01105, which is a Core i3-10105. https://www.tomshardware.com/news/chinas-powerstar-cpu-seemingly-confirmed-as-intel-silicon-via-geekbench
|
# ? Dec 14, 2023 15:38 |
|
|
# ? Jun 13, 2024 05:06 |
|
Twerk from Home posted:Via is back, baby. I'm really surprised how dramatically they seem to have caught up, too: Curious since they're claiming chiplet design, which afaik Intel is still only doing monolithic dies. Though they could also just be saying that and it not actually be the case. I'd like to see one delidded so we can get a peek at what's under the IHS.
|
# ? Dec 14, 2023 16:02 |