What's your latest tech purchase?

[Mussels]

Ended up sending it back based on the feedback from their support team.

so they can resell it to someone else, who will use it for their 25Mb DSL

 

[fusseli]

Got a 7900xtx. No regerts.

 

[Mussels]

Got a 7900xtx. No regerts.

I hope that was intentional, because i giggled so much I fell over.

I made a "no egrets" coffee mug for a friend who does bird photography - cant find the original but the image was like this

 

[sam_86314]

Got a new boot SSD for my main PC.

Went with a 2TB WD SN850X for $110 before tax. This continues my trend of doubling the size of my boot drive for the same price; my first NVME SSD was a 500GB Samsung 970 EVO for $120 in 2018, then I got a 1TB WD SN750 for $120 in 2021, and then this for $110.

While I'm not going to pull the label off to identify the chips, the SSD Database says it uses 112-layer TLC from Toshiba/Kioxia, has 2GB of Micron DDR4-3200 as a cache, and uses a custom controller from WD/SanDisk.

I was seriously considering going with a 2TB Fanxiang S770 out of curiosity. That drive is regularly on sale for like $86, but I've heard that the Innogrit controller it uses has lots of issues. I've had excellent experiences with WD SSDs thus far, and SSD prices are really good right now, so this seemed like the better option.

I ended up doing a fresh install of Windows since my old install had tons of problems. As expected, it was a royal pain in the ass, but after messing around with partitions and EFI nonsense, I got it working.

It performs great in synthetics, especially compared to my previous drive. It runs very hot though, reaching 79C while benchmarking (the SN750 got up to around 67C I think). Clearly, the built-in SSD heatsinks on my motherboard are woefully inadequate.

The next step is to finish setting up Windows and then install Linux alongside it (I'm going to give OpenSUSE Tumbleweed a try instead of the Debian-based stuff I usually use).

 

[lexluthermiester]

you'll either get hard crashes 'out of memory' or have windows automatically expand it, if it's too small

With 32GB of RAM and a 6GB pagefile?!? Never gonna happen...

hiberfile is the space eater that matches your RAM size

I completely disable that crap. Hibernation has no place on a modern PC.

 

[nomdeplume]

It performs great in synthetics, especially compared to my previous drive.

Out of curiosity, did you just open CDM and use the normal bench or set it to the NVMe specific one?

 

[sam_86314]

Out of curiosity, did you just open CDM and use the normal bench or set it to the NVMe specific one?

I set it to NVME.

Here's another speed test after I've gotten Windows all set up (CDM set to NVMe).

Yes, I've already written over 1TB to the drive. Stupid Windows bootloader nonsense...

 

[Mussels]

With 32GB of RAM and a 6GB pagefile?!? Never gonna happen...


I completely disable that crap. Hibernation has no place on a modern PC.

Yes, because some apps allocate large amounts of space to the page file directly regardless of what they use.
pops in in gaming groups all the time, people misunderstand the memory errors and think its RAM related, not virtual memory

 

[Super XP]

Forgot to mention. Also got this too. To add more M.2 drives. This thing is super fast too.

ASUS Hyper M.2 X16 PCIe 4.0 X4 Expansion Card
* Supports 4 NVMe M.2 (2242/2260/2280/22110) up to 256Gbps for AMD 3rd Ryzen sTRX40, AM4 Socket and Intel VROC NVMe Raid

 

[Ruru]

A 90 degree adapter for the internal USB3 connector. The connector is under the GPU cooler so there was no way to put that cable in without getting such an adapter.

 

[QuietBob]

It runs very hot though, reaching 79C while benchmarking ... Clearly, the built-in SSD heatsinks on my motherboard are woefully inadequate.

That does seem a bit too hot, unless there was no proper contact between the drive and the MB heatsink. Considering that you could see even higher temps in games -- when the drive gets heated up by the 240 W GPU -- you may want to install a third party cooling solution. I just ran the CDM NVME benchmark in a 27 C room on my 1 TB 850X with an aftermarket passive heatsink and got this:

Max temp was 49 C during read tests, and 52 C during writes. Idle is 40 C.

 

[sam_86314]

That does seem a bit too hot, unless there was no proper contact between the drive and the MB heatsink. Considering that you could see even higher temps in games -- when the drive gets heated up by the 240 W GPU -- you may want to install a third party cooling solution. I just ran the CDM NVME benchmark in a 27 C room on my 1 TB 850X with an aftermarket passive heatsink and got this:

View attachment 306614

Max temp was 49 C during read tests, and 52 C during writes. Idle is 40 C.

That's most likely it. I noticed that the heatsink came off very easily. Could it be a design flaw with my board; that the heatsink sits slightly too high?

EDIT: I've ordered a 3mm-thick thermal pad to see if that will help. It's supposed to arrive tomorrow.

 

[Deleted member 229121]

So in lieu of recent events I went with Plan C, keep the old g-sync display and go for a cheap upgrade.

Ordered a 4070 for $420 after tax w/ some gift card shuffling.
Selling the 3060 Ti to someone local for $200 as well.

I'll just wait a bit longer to make the leap to 4K, it's all good - the displays should be better/cheaper then too.

 

[sam_86314]

That does seem a bit too hot, unless there was no proper contact between the drive and the MB heatsink. Considering that you could see even higher temps in games -- when the drive gets heated up by the 240 W GPU -- you may want to install a third party cooling solution. I just ran the CDM NVME benchmark in a 27 C room on my 1 TB 850X with an aftermarket passive heatsink and got this:

Max temp was 49 C during read tests, and 52 C during writes. Idle is 40 C.

Alright, I grabbed one of the thin thermal pads that came with my Fanxiang S660 and stuck that on top of the pad that came with my board's SSD heatsink.

'

Thermal pad thickness was definitely the issue. It now only gets up to the mid-50s. It also now idles around 40C (23C ambient) instead of 55C.

Stacking thermal pads might not be ideal, but it works.

 

[tabascosauz]

Alright, I grabbed one of the thin thermal pads that came with my Fanxiang S660 and stuck that on top of the pad that came with my board's SSD heatsink.

Thermal pad thickness was definitely the issue. It now only gets up to the mid-50s. It also now idles around 40C (23C ambient) instead of 55C.

Stacking thermal pads might not be ideal, but it works.

Seems about right. WD NVMes generally run a bit warm ever since the original Black. As long as you make contact with the controller package where the temp sensor is, should be fine. I have a bit of an unfair airflow advantage with SO-DIMM.2.

MSI is a fan of stacking thermal pads - my Unify-X had two layers of thermal pads that you could remove/add at will depending on whether your NVMe was single or double sided. Didn't really change temps for better or worse.

I'm debating whether I should replace my 1TB SN750 now that things are so cheap. I paid more than double what you paid for the SN850X when it launched.

 

[lexluthermiester]

Yes, because some apps allocate large amounts of space to the page file directly regardless of what they use.
pops in in gaming groups all the time, people misunderstand the memory errors and think its RAM related, not virtual memory

I've never once seen such a problem. As far as my experience goes, it might as well be a total myth.

 

[Mussels]

Forgot to mention. Also got this too. To add more M.2 drives. This thing is super fast too.

ASUS Hyper M.2 X16 PCIe 4.0 X4 Expansion Card
* Supports 4 NVMe M.2 (2242/2260/2280/22110) up to 256Gbps for AMD 3rd Ryzen sTRX40, AM4 Socket and Intel VROC NVMe Raid

The amount of people on the FB PCMR groups buying those only to find out "oh wait this needs my 16x slot and now i cant have a GPU?"
even using the secondary slot with bifurcation, you can only use two drives on it.

I've never once seen such a problem. As far as my experience goes, it might as well be a total myth.

because normally, the OS will invisibly create/expand the file. These people disable it or shrink it and run their C: drives full, so theres no room for that expansion so it goes crashy crashy.


its common enough it comes up in google searches

 

[nomdeplume]

Alright, I grabbed one of the thin thermal pads that came with my Fanxiang S660 and stuck that on top of the pad that came with my board's SSD heatsink.

Nice, that ended up being an easy solution.

 

[lexluthermiester]

because normally, the OS will invisibly create/expand the file.

No what I meant was I have never seen that error on systems I have built and custom configured. It just doesn't happen. Ever.

These people disable it

AH HA! That's what can be causing the error for some programs.

or shrink it

Depends on how far. 512MB, 1GB or 2GB on a system with only 8GB? Yeah crashes might happen. But 16GB or more system RAM with 4GB or more pagefile? Never gonna happen. Ever.

So 32GB system RAM and a 6GB pagefile is perfectly fine for any operating scenario.

 

[Mussels]

No what I meant was I have never seen that error on systems I have built and custom configured. It just doesn't happen. Ever.

because you're not running out of space on the C: drive...

 

[lexluthermiester]

because you're not running out of space on the C: drive...

That's not a factor.

 

[Mussels]

That's not a factor.

It's 100% the thing that triggers the problem.
I don't understand what you're thinking right now.

 

[lexluthermiester]

It's 100% the thing that triggers the problem.

In systems with pagefiles smaller than 4GB and system RAM less than 8GB, maybe.

I don't understand what you're thinking right now.

My point is that with >4GB pagefile and >16GB of system RAM, it's not going happen. This is the advice I gave to @nomdeplume earlier. They have 32GB of system RAM. A locked 6GB pagefile will prevent drive access thrashing and will always be enough for Windows and apps to run in.

I think I wasn't being as clear as was needed with the definitions and context. My bad Mussels.

 

[TheLostSwede]

Ended up with this thing instead, it's a generic thing branded by some Taiwanese company, but at least they point out it's the RTL8156B chip and it seems to be working as it should.

C:\iperf>iperf3 -c 192.168.1.58
Connecting to host 192.168.1.58, port 5201
[  4] local 192.168.1.83 port 50196 connected to 192.168.1.58 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   278 MBytes  2.33 Gbits/sec
[  4]   1.00-2.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   2.00-3.00   sec   283 MBytes  2.37 Gbits/sec
[  4]   3.00-4.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   4.00-5.00   sec   282 MBytes  2.36 Gbits/sec
[  4]   5.00-6.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   6.00-7.00   sec   276 MBytes  2.32 Gbits/sec
[  4]   7.00-8.00   sec   271 MBytes  2.27 Gbits/sec
[  4]   8.00-9.00   sec   276 MBytes  2.32 Gbits/sec
[  4]   9.00-10.00  sec   281 MBytes  2.36 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  2.73 GBytes  2.34 Gbits/sec                  sender
[  4]   0.00-10.00  sec  2.73 GBytes  2.34 Gbits/sec                  receiver

iperf Done.

C:\iperf>iperf3 -c 192.168.1.58 -R
Connecting to host 192.168.1.58, port 5201
Reverse mode, remote host 192.168.1.58 is sending
[  4] local 192.168.1.83 port 50222 connected to 192.168.1.58 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   262 MBytes  2.20 Gbits/sec
[  4]   1.00-2.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   2.00-3.00   sec   277 MBytes  2.33 Gbits/sec
[  4]   3.00-4.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   4.00-5.00   sec   282 MBytes  2.36 Gbits/sec
[  4]   5.00-6.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   6.00-7.00   sec   276 MBytes  2.31 Gbits/sec
[  4]   7.00-8.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   8.00-9.00   sec   281 MBytes  2.36 Gbits/sec
[  4]   9.00-10.00  sec   282 MBytes  2.36 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  2.72 GBytes  2.34 Gbits/sec                  sender
[  4]   0.00-10.00  sec  2.72 GBytes  2.34 Gbits/sec                  receiver

iperf Done.

Now this was tested against an Intel i-225V, as for some reason, I'm having major issues with my Aquantia 10 Gbps cards and pretty much anything I connect to them. It's fine to send data to the Aquantia card, but receiving data from them ends up at no more than 600 Mbps, which is just insane. Here's the kicker though, Aquantia to Aquantia works just fine, so it seems like Marvel has introduced some kind of driver bug, which will be impossible to have fixed, since they don't do end user support, unlike Aquantia.

I know this is somewhat the wrong thread for this, but the previous adapter I bought, might just have been fine. Oh well...

 

[lexluthermiester]

Ended up with this thing instead, it's a generic thing branded by some Taiwanese company, but at least they point out it's the RTL8156B chip and it seems to be working as it should.

C:\iperf>iperf3 -c 192.168.1.58
Connecting to host 192.168.1.58, port 5201
[  4] local 192.168.1.83 port 50196 connected to 192.168.1.58 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   278 MBytes  2.33 Gbits/sec
[  4]   1.00-2.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   2.00-3.00   sec   283 MBytes  2.37 Gbits/sec
[  4]   3.00-4.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   4.00-5.00   sec   282 MBytes  2.36 Gbits/sec
[  4]   5.00-6.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   6.00-7.00   sec   276 MBytes  2.32 Gbits/sec
[  4]   7.00-8.00   sec   271 MBytes  2.27 Gbits/sec
[  4]   8.00-9.00   sec   276 MBytes  2.32 Gbits/sec
[  4]   9.00-10.00  sec   281 MBytes  2.36 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  2.73 GBytes  2.34 Gbits/sec                  sender
[  4]   0.00-10.00  sec  2.73 GBytes  2.34 Gbits/sec                  receiver

iperf Done.

C:\iperf>iperf3 -c 192.168.1.58 -R
Connecting to host 192.168.1.58, port 5201
Reverse mode, remote host 192.168.1.58 is sending
[  4] local 192.168.1.83 port 50222 connected to 192.168.1.58 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   262 MBytes  2.20 Gbits/sec
[  4]   1.00-2.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   2.00-3.00   sec   277 MBytes  2.33 Gbits/sec
[  4]   3.00-4.00   sec   280 MBytes  2.35 Gbits/sec
[  4]   4.00-5.00   sec   282 MBytes  2.36 Gbits/sec
[  4]   5.00-6.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   6.00-7.00   sec   276 MBytes  2.31 Gbits/sec
[  4]   7.00-8.00   sec   282 MBytes  2.37 Gbits/sec
[  4]   8.00-9.00   sec   281 MBytes  2.36 Gbits/sec
[  4]   9.00-10.00  sec   282 MBytes  2.36 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  2.72 GBytes  2.34 Gbits/sec                  sender
[  4]   0.00-10.00  sec  2.72 GBytes  2.34 Gbits/sec                  receiver

iperf Done.

Now this was tested against an Intel i-225V, as for some reason, I'm having major issues with my Aquantia 10 Gbps cards and pretty much anything I connect to them. It's fine to send data to the Aquantia card, but receiving data from them ends up at no more than 600 Mbps, which is just insane. Here's the kicker though, Aquantia to Aquantia works just fine, so it seems like Marvell has introduced some kind of driver bug, which will be impossible to have fixed, since they don't do end user support, unlike Aquantia.

I know this is somewhat the wrong thread for this, but the previous adapter I bought, might just have been fine. Oh well...

2.3ishGbps? That's not bad for an inexpensive pluggable network card.