ChalkChalkson

I'll take a look again Yeah GN testing suggested 4x vs 2x made about 8% performance difference on CL14 3200, the price difference was like 5-10€, so that's probably very cost effective, similar story for CL14 vs CL16... Looks like a fair amount of the real cost of Ryzen is in the better RAM you need to use it most efficiently Thanks a lot!

ffs I can't believe I screwed this up twice... I'm in bed now, but here is an imgur link to the parts list I'll edit it tomorrow with the PC partpicker

Oh yeah here is a PC partpicker list. Though the price on the memory is slightly above what mindfactory shows.

Hey, A friend of mine asked me for help in upgrading the PC I built for him in 2014. Due to the GPU shortage he'll probably keep rocking his 970 for a while, but this is the new stuff which I'll probably suggest to him. Besides playing games (though mostly stuff like LoL and Anno) he uses his PC for running scripts he writes for his MSc project. I know his code though, he rarely multi-threads The part I'm most unsure about is the RAM MoBo pairing. GN's testing suggested to me that 4x 3200 14-14-14 memory is still very cost effective (3600 cl14 would be at least 20% more expensive if I checked right) - though I could have interpreted that completely wrong. Both the board and the ram kit are rated for 3200, but the QVL of the motherboard has it listed as such with "native" 2133 speed, while some other 3200 kits are listed as native 3200. Does that mean I shouldn't expect them to work together cleanly at 3200, or just that I actually have to enable XMP? Speaking of XMP, this kit doesn't have the "v" in that column either which slightly concerns me.. I know this is a bit more in depth on the questions than these posts typically are, but if you know the answer to either of those questions, or spotted anything else weird about the component list, please let me know And thanks for taking the time to read this either way!

I terminated the ends pretty recently (like 3 weeks ago? the issue is a bit older) when I did I checked continuity with a multimeter - the cables are all in tact and wired up in matching ways. If you asked me whether I wired it matching or crosswired (A-A or A-B) I couldn't tell you though. Is that a potential cause?

Appears that the link into the basement is limited to 100Mb - or at least plugging my laptop into that switch only allows me to transfer to it from my PC at 100Mb, while transfering to it when plugged into the ground floor router works at Gb. This is hella annoying the cable is pretty long and passes through multiple concrete walls / ceilings without cable channels O.o

on it

Yeah it's not a switch it's a router switch AP combo. But I use it as a switch + AP, got some for cheap a while ago. What do you mean by trunk cables? The cables that go from the devices to the switches and switches to the patch panels? Those are all Gb. One of the first things I checked^^

Task manager network tab (what I show above) shows bit. The transfer window (what I showed there wasn't a windows file transfer so I can't take the corresponding snapshot) does show MB. Windows transfers when I do them tend to clock in at ~10MB/s

The two switches are identical, they're Asus AC1900s which are gigabit switches (again I managed to get gigabit speeds to different point in the network through either of them)

router at ground level, cat6 patch running to a switch on the first floor and basement each, my PC and NAS are plugged into those respectively

Jup, you can see in the screenshot that they both show that they are hooked up at gigabit. And again, both managed to get 200Mb to my router

I'm talking about writes inside the network here, not my connection to the internet. Though when I had 200Mb/s internet I got those speeds on both my NAS and my PC

MSI Z170 titanium on my PC and uh... some AsRock B365 on the NAS. I don't have a great way to test it from all systems, but when I'm very close in the network to the NAS I can write to it at Gb. And my PC can write to other PCs that are close in the network at Gb, too I think.

Huh. Windows task manager shows 90MBit/s. No idea whats wrong with the picture - shows it fine on my end

Mb. I'm such an idiot - I even took a screenshot and forgot to put it in the post O.o

Hi! I recently redid my homes networking, cat6 cableing, all gigabit switches. All my devices (at least those that support it) show gigabit connections. But when I write a large file from my computer to my NAS the speed is limited to ~100Mb/s significantly below the respective read and write speeds. Any ideas what might be going on?

Not really needed for "direct die cooling" I'd guess. If you think about it, the hotplate of the cooler is basically an IHS in and of itself. Exactly what I was thinking If you use only the pads it'd take forever to do though (so thinn). So my best guess would be using alternating layers of this and doublesided tape.

I mean over the thermal paste you often have a relevant temperature drop. Guess it's just that I compare this to "good" metal to metal contact and you compare it to thermal paste.

huh, dont have much experience with that particular chip, but the temperature seemed high for a dry mount with adequate pressure. This basically just shows that these pads are more a replacement for the IHS than for the thermal pad.

Nope there is more to it. If all that was going on was the shitty elasticity, you'd see performance very similar to that of a dry mount (which isn't that awful). The physics of pyrolitic carbon is such that is only conducts in one plane (thread where I explained how and why after the floatplane release). Or as figgyc said: Since it's in the description of the product I'd assume @AlexTheGreatish was aware. Pretty hyped to see what they are planning with it. That could work. In the thread I linked earlier we also chatted for a second about thermal compounds that use small fibers of carbon to increase conductivity. Blending would be considerably worse though since it produces flakes. The performance you'd get would essentially be the average of the conductivities in all directions which is to say you put a lot of effort into getting slightly worse graphite. Maybe if you could find a way of getting ~1-10µm thin strips that are ~100µm long, then suspending them in a silicone rubber of some sort and aligning them to poke up vertically through the use of an electrostatic field....

Wouldn't you expect a suspension to have an even distribution of orientations? I'd love to see some scattering graphs of these TIMs. EDIT: just looked at the SEM pictures in that review. Pretty interesting! I have to say when I said IC Diamond I was thinking about the thermal paste with diamond particles. Would also like to know what the structure of the carbon fibers in this material is, expecting somehting polycrystaline. Again, I what I'd love most would be to see someone take those pads and pastes and do some scattering experiments. If only I had gone down the route of applied physics, could have been a fun BSC thesis

In the ribbon cable example you'd apply the mounting pressure in the Z direction, so there it should be fine. For a theoretical compound thermal pad where the X direction goes from the IHS to the cooler, you'd hope that the matrix gives the needed mechanical properties. IIRC that's what they do with graphene fragments in epoxy. But this is super outside my wheel house. You'd need to ask a composite materials person. Well, these high tech carbon based materials have stupidly high conductivities, so the increased resistence where the contact might otherwise be metal to metal could be negligable. And where the pad actually fills gaps it is a lot better than even diamond particles in suspension (IC diamond). But let's be honest here, pyrolitic carbon composites are kinda overengineering a very simple problem. That's why I kinda like the idea with the ribbon cable, at leat that is something where the high conductivity actually gives a meaningful advantage over metal. In fact, the theoretical performance is so high, it should even beat heat pipes...

TLDR; Say one happened to have a pad of pyrolitic carbon with the graphene sheets being coplanar to the pad. Then maybe it would be cool to do something with it where the conductivity in the direction the pad is actually conductive in mattered and not in the direction the conductivity sucks. For example making a heat conducting ribbon cable to cool a chip remotely. I don't want to spoil anything, so let's just say @AlexTheGreatish was at it again with cool cooling stuff. Why Graphite Conductivity is Anisotropic Super short background on what graphene, graphite and pyrolitic carbon are (sheets and how they stack): So, what is all the fuzz about? Well the graphene sheets are really cool. Because of they are conjugated systems (read: because their chemistry is cool) electrons can move freely through the entire sheet - almost exactly like a metal! You might have heard that moving charges is what conducts electricity, so you would assume that graphene sheets are really good at conducting heat along their surface. And you'd be correct! Now, much more interesting for this subforum, the "charge" for heat conduction is just energy, and basically anything can transmit energy, including electrons. That's why graphene (and metals) are so good at conducting both. That's all nice, but graphene is one of those sci-fi materials that, may not be the stuff of dreams anymore, but are very much still the stuff of research papers, not commodity items. Graphite however is. The issue is that electrons have a much harder time to jump between layers than they have moving along them. This means, if you have a stack of graphene sheets oriented in the X-Y plane, you will measure excellent conductivity in those directions and bad conductivity in the Z direction. In ordinary graphite, by the time the heat or current has traveled through the material, it will have passed many zones, in some it will have traveled along the sheets in some it had to travel across them, So in effect you will measure a conductivity that is something inbetween. With pyrolitic carbon however all of them are aligned. So the pad itself will have an X and Y direction where the heat travels really well and a direction where it travels really badly. So let's wildly imagine having such a pad is large in the X and Y directions and thin in the Z direction. You would expect it to be so good at conducting in the X-Y plane that touching it at two far away points in that plane would be enough to melt your way through an ice cube. You would also expect it to be really electrically conductive along that surface. But, it would really suck as a thermal conductor in the Z direction. (just making things up tough) When something varies with the orientation it's called anisotropic (an means not, iso equal, tropos means place or path/way or something like that). The anisotropy of graphite (meaning just one sheet stack here, because we physicists like to pretend cows are spheres....) is very well known. Here is a paper in nature from the 30s (lower left hand side). In fact it is pretty much *the* example of macroscopic anisotropy resulting from crystal structures for intro to condensed matter physics courses If It's a Shit Thermal Pad, What Can You Do With It? Let's say for some reason you bought some pyrolitic carbon on @LinusTech's credit card and wanted to use it for somehting that works to make him happy.... A thing you could try is to cut a strip as wide as a chip and a couple times as long. Glue it onto the chip and stick a cooler on the other end of the strip. If the thermal conductivity is really above 500W/mK then cooling a chip through a 10cm x 5cm x 0.1mm ribbon cable should "only" add like 1K per 5W. So hypothetically, cooling something like a Pentium G5XXX with a decent cooler and the ribbon cable could work. 500W/mK is around 30% of the theoretical performance of pyrolitic carbon, so there is a chance.... (Plus it'd look dope through a thermal camera). If the thing performs at closer to 60% of theoretical performance, maybe you can even make the strip long enough to put the cooler on the back.... Even if the obect is never to be seen again. I'd love to hear if the electric conductivity through the pad is noticably worse, since that would give me more reason to believe that this anisotropy was what caused the poor result. If It Ain't Graphene Then What Is the Sci-Fi Material For Thermal Pads? The ideal themal pad in not electrically conductive along the X-Y plane and has awesome thermal conductivity in the Z direction. It also needs to be flexible. From the discussion above it should be cleat that it would be great to have a 1D equivalent for graphene. This material exists - and much to Riley's enjoyment - it's carbon nano tubes. So my guess for the best thermal pad would be carbon nanotubes all aligned in the Z direction and stuck together with silicone rubber or something like that. Thinking about it, those glue coated graphene sheets are basically the exact opposite of the ideal thermal pad... funny. You folks might actually be able to simulate this with the materials you have. Cut the sheets into thin 5cm long strips and glue them together into a 5x5cm block, then slice a thing layer off with a very sharp razor. That way the good conduction direction is actually aligned with the direction you need it to conduct in. Not sure if the fabrication works, I'm a theoretical physicist, someone who knows how to actually manufacture stuff like @AlexTheGreatish... Though with the 16µm sheets you have that'd be like 3000 layers. But, if you for example alternated these sheets with 0.1mm double sided tape, that'd go down to a couple hundred layers. And if they do actually conduct at 1000W/mK along the sheet, the resulting composite would have a conductivity of around 80-100W/mK in the relevant direction without being electrically conductive in that relevant direction... Maybe worth a try on a small scale.

I mean my old nas ran on a consumer mobo and I had the igpu running a gui and a dedicated gpu passed through to a vm