Corsair Vengeance RGB CUDIMM DDR5-8800 48 GB CL42

[Dr. Dro]

42-54-54-141-195 are not that good in terms of timings.

Team Group has the same 8800MHz memory in 42-54-54-96

You can get CL40 if you drop to 8600MHz.

The speed is not everything, the timing has to be in line with the speed. That's why in the charts the 7600MHz memory is beating this kit because of better timings. I personally would get 8200MHz memory at 38-49-49-84

Subs tend to be heavily dependent on the IC type employed, I suppose. Considered this isn't exactly marketed as a high performance kit, I guess they've slacked here and use chips with looser binning, you may get a kit that tightens well or another one that does not.

Pricing is bad for a midrange, that's for sure though.

 

[phanbuey]

this seems like it doesn't actually do anything for performance.

 

[ir_cow]

this seems like it doesn't actually do anything for performance.

Correct. The bandwidth is there but Intel 200 series truly is latency dependent for performance. We won't see the benefits until AMD supports CUDIMMs.

That, or you take things into your own hands and manually change the ring cache and D2D.

My point is, when you go so far outside the specs, you're asking for trouble.

That's is true for anything outside of the CPU officially supported memory speeds. For you, the AMD X3D is perfect.

 

[Watermelonn]

My point is, when you go so far outside the specs, you're asking for trouble.

DRAM ICs generally have great voltage tolerance. None of the DDR5 ICs have been proven to die from any voltage with proper cooling (though some scale negatively on timings/frequency), the only DDR4 that has been proven to die from voltage are hynix ICs before 8gbit DJR (with 1.4-1.6v being the most I'd consider dailying depending on the IC) and no DDR3 IC has been proven to die from voltage (tough the later hynix may be mildly sus at like 2V+).

At this point, Hynix 16gbit A-die and 24gbit M-die have been extensively tested at 1.8v and higher and these voltages would be perfectly fine to daily.

 

[dgianstefani]

DRAM ICs generally have great voltage tolerance. None of the DDR5 ICs have been proven to die from any voltage with proper cooling (though some scale negatively on timings/frequency), the only DDR4 that has been proven to die from voltage are hynix ICs before 8gbit DJR (with 1.4-1.6v being the most I'd consider dailying depending on the IC) and no DDR3 IC has been proven to die from voltage (tough the later hynix may be mildly sus at like 2V+).

At this point, Hynix 16gbit A-die and 24gbit M-die have been extensively tested at 1.8v and higher and these voltages would be perfectly fine to daily.

The issue is the relationship between memory controller and DRAM voltage.

 

[Watermelonn]

The issue is the relationship between memory controller and DRAM voltage.

I can't speak for DDR5 as it's not a standard I'm familiar with (tough afaik all IMCs have been tested rather extensively at 2v), but for DDR4 IMCs, none die from vdimms below 2v. Skylake-based archs I'd consider anything above that sus for daily (not that there's any point to running that for daily past memeing), as 2.25v has only been tested for short times. DDR3 is different as AMD Kaveri-based APUs will quickly die from 1.8v, though 1.65v should be considered safe as AMD marketed kaveri-compatible kits at that voltage. Sandy bridge is also known to degrade with 2v.

The TL;DR is that you're working with rather low current components, so outside of a few edge cases, what might sound like dangerously high voltages for other components like CPU cores is perfectly safe for DRAM and IMCs. (also the IMC itself doesn't run at the same voltage, only the PHY)

 

[ir_cow]

Put 2.0v into SK Hynix DDR5 and it will not survive long. Micron dies at 1.4v sometimes lol.

It also doesn't scale much past 1.6v anyways for SK Hynix because your fighting IC temperatures first and foremost.

 

[Watermelonn]

Put 2.0v into SK Hynix DDR5 and it will not survive long. Micron dies at 1.4v sometimes lol.

It also doesn't scale much past 1.6v anyways for SK Hynix because your fighting IC temperatures first and foremost.

I'm pretty sure M16D CPOC has been tested above 1.4v VDD/VDDQ, by "die" do you just mean that unbinned M16D/M32B negatively scales harshly above 1.3v VDD?

 

[ir_cow]

@watermolonn like no longer functioning. That kind of dead. DDR4 didn't do that well either with high voltage if I recall.

You can do whatever you like with memory voltages, but I highly suggest against going to the extremes for anyone else reading this.

Say you have good enough cooling for 2.0v what does that really accomplish?

 

[Watermelonn]

Dunno about DDR5 since I don't bench that and the micron ICs are very poorly documented. But for DDR4, M8E and M16B (which are like, the only two relevant ICs) have been rather extensively tested ambient up to 1.6-1.7v, beyond that they'd be rather hard to cool anyways since they're inefficient af. Besides some of their early stuff showing very strong negative scaling with voltage, most of it should be fine?

Which micron IC did you manage to kill at 1.4v, VDD or VDDQ?

 

[bug]

DRAM ICs generally have great voltage tolerance. None of the DDR5 ICs have been proven to die from any voltage with proper cooling (though some scale negatively on timings/frequency), the only DDR4 that has been proven to die from voltage are hynix ICs before 8gbit DJR (with 1.4-1.6v being the most I'd consider dailying depending on the IC) and no DDR3 IC has been proven to die from voltage (tough the later hynix may be mildly sus at like 2V+).

At this point, Hynix 16gbit A-die and 24gbit M-die have been extensively tested at 1.8v and higher and these voltages would be perfectly fine to daily.

Nice argument there. How do you prove a DRAM IC has died because of overvoltage? Because surely users won't RMA with a "died whan I was running it @2V" label on it.
Also, what do you define as "proper cooling" considering most RAM sticks come with a simple heatspreader that relies on case airflow?

Fwiw, I wasn't saying RAM will automatically die when overvolted. I was just saying when you go more than 5-10% out of standardized values, changes for you RAM stick to not work as advertised shoot significantly up.

 

[Watermelonn]

Nice argument there. How do you prove a DRAM IC has died because of overvoltage?

Notice how I worded it as "proven to die" rather than outright stating that the IC is "known to work fine at X voltage". It's mostly based on user reports with some extra work to weed out ICs that just suffer from random deaths.

Also, what do you define as "proper cooling" considering most RAM sticks come with a simple heatspreader that relies on case airflow?

You really gotta be trying to push newer DRAM ICs beyond 70-80°C. Ideally it's a fan blowing over the bare dimms, and less ideally a fan blowing over the heatspreaders.

Fwiw, I wasn't saying RAM will automatically die when overvolted. I was just saying when you go more than 5-10% out of standardized values, changes for you RAM stick to not work as advertised shoot significantly up.

DRAM manufacturers actually used to claim voltages beyond JEDEC in their specs, Micron was rather generous and some of their DDR3 ICs were specced up to the optional 95°C and a significantly increased maximum voltage rating of 1.975v (spectek special bin V80A, can't be bothered to check other specs but it shouldn't be the only one). It seems rather likely that DRAM manufacturers communicate with 3rd party DIMM manufacturers for some specs, and thus you can trust the highest voltage XMP ICs appear in as an absolute minimum for the maximum safe voltage for long term use. This could already be significantly higher than spec, as DDR4 Hynix 8gbit DJR was shipped in 1.6v (or 33% over JEDEC) XMP kits.

 

[ir_cow]

DDR4 Hynix DJR was just fine at 1.6v. necessary if you wanted to hit 5333 MT. CJR... Not so much. Seen enough reddit posts were 1.45v was enough to off it.

I remember buying Samsung B-Die was reading about how it has no problems with 1.8v. After 2 months, one DIMM was dead. All for what? 4000 CL14 and a loss of money?

There is no universal "okay" voltage. It various per IC manufacturer. longevity and temperature goes hand in hand. At 80c, you are lucky even XMP is stable.

 

[Watermelonn]

CJR... Not so much.

Yea, CJR is one of the worse early DDR4 Hynix ICs, I'd probably consider a few others fine at 1.5v but I'd daily CJR at 1.4v.

I remember buying Samsung B-Die was reading about how it has no problems with 1.8v. After 2 months, one DIMM was dead. All for what? 4000 CL14 and a loss of money?

This just sounds like skill issue, S8B has been extensively benched up to 2.2v with maxmem at ambient, there are a few reports of deaths at 17-1.8v that I'd rather attribute to defective chips or other DIMM-related issues than the design itself dying at that voltage.
Running S8B at 1.8v for daily is odd anyways since you'd be suffering from rather harsh tRCD rollover.