Comments by "eDoc2020" (@eDoc2020) on "Gamers Nexus" channel.

I feel compelled to mention it here. The two Linuses have different names: Torvalds (Linux guy) is "Lee-nus" Sebastian (drop things guy) is "Lie-nus"
11
Most servers don't have iGPUs and instead use a variant of the Matrox G200 as their video card. It's not the latest and greatest card by any means but the company is still around.
5
Technically Microsoft thinks 10 comes after 6.2.
3
@PainterVierax I'm absolutely certain my HP Gen8 uses Matrox. It's embedded in the BMC and the BMC itself is able to capture the display output to provide a virtual KVM console over the network. The ARM chip in the BMC doesn't have a video output. While some more specialized servers are truly headless all "standard" x86 servers still have a video chip. It appears Dells also use Matrox graphics. If you look at the downloads page for the latest Poweredge R740 you'll find the Matrox Video Driver, currently at version 4.4.1.3 released on 09 Dec 2019. It has Dell's driver ID 1dvtp and you can see the huge list of servers it is used with.
3
@bobprivate8575 You're pretty close. A regular cheapo 120mm fan from my box of extra fans (Yate Loon D12SH-12) is rated at 0.3A (still not much) and the fan from a BTX Dell is rated 1.4A. Drives are also a bit more. Spinup power is more like 30W (actually less than you said) but idle power can be a bit over 7W and in use power is close to 12W. As you said, this still isn't going to be a huge issue. Good PSUs can handle a bit of overload for a while, there's sometimes even a listed spec for this. An old prebuilt I have has a 12V rail rated 4A continuous or 7A for 15 seconds maximum. I wonder why we don't see ratings like that on modern PSUs.
2
@roboticvenom1935 High performance fans can use a lot of power. I have some powerful 60mm fans rated at 3.3 amps. They actually only draw 1.8a (20w) full speed but that should still be a consideration. Granted a gaming PC is unlikely to have fans like that that but a regular 120mm fan can probably use 6w. Multiply that by several fans and it starts to add up. In terms of IO all your drives will use some power and everything powered from your USB ports of course also comes from the PSU.
2
@PainterVierax If you're interested, here's the link for the driver on Dell's site I mentioned. I'm doing this as a separate comment because I'm not sure if links in comments are allowed here. https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=1dvtp
2
@scottyhaines4226 Servers care a lot about drive speed, especially random access speed where SSDs totally win. In the old days you'd often need an array of lots of drives not for the capacity but for the random IOPS. With SSDs you don't need as many drives and the drives use much less power. Power usage / heat generation is an important consideration in the datacenter because they not only need to power the equipment but also need to pay for cooling and air conditioning. AFAIK in the server world spinning disks are pretty much dead for general purpose use (vs bulk storage). SSDs let you handle more clients in fewer servers, offering great cost savings.
2
@bobprivate8575 The Dell fan I'm talking about from a Dimension E521 was actually made by NMB in China. I have another one, probably from an Optiplex, which may be a Delta but I can't find it at the moment. Fortunately I know I won't accidentally plug it into a 1A mobo header because it has a different connector. I also happen to have an HP server with 60mm Delta fans (Chinese according to pictures I can find online) rated 3.3 amps. When they are spinning full speed the system is indeed very loud (even at 35% it gets irritating after a few minutes) but I measured and they only use 1.8 amps. This server also originally shipped with twelve 15000 rpm 3.5" drives which would've used around 180 watts in use. But yeah, neither of these are going to be found in an ordinary PC. Also, 100 linear feet/minute isn't that much. A 120mm fan is about .12 square feet so 100 ft/min would be 12 CFM. For comparison, a Noctua NF-A12 is a max of around 60CFM.
1
@bobprivate8575 I'm sure about those numbers. The datasheet for the Hitachi 15k600 says 16.6w typical operating power and for the 600gb capacity. Drag forces are proportional to the square of the velocity so I'd expect 2.25 times the spindle motor power when going from 10k to 15k rpm. With fan measurements the smaller the fan area the more velocity if the CFM stays the same. I think airspeed of a fan doesn't matter unless you need to cross the bridge of death. For most purposes CFM (or m3/hr for metric folks) is probably the more useful metric. 100 CFM sounds plausible for a high power 92mm fan. On second thought my 60mm ones might be rated around 150 CFM, so 100CFM seems low. Either way, it's completely irrelevant for sizing the PSU on a gaming PC.
1
I'm as upset with Microsoft dropping support for middle-aged CPUs as much as the next person, but where was the outrage when Windows 8 did the same thing?
1
I'm sure they could make one if they wanted to. P4 family chips could do 4GHz back in the day without much difficulty and semiconductor fabrication has gotten better since them. Of course, who would ever buy one if you could get a 5GHz C2D?
1
That shouldn't really matter, they should be using cookies to tell separate users apart. Having said that, it may have an impact on the issue but it's not the issue.
1
Fortunately SSDs start at lower prices than HDDs (obviously with less capacity, though)
1
With some little soldering the LX can actually go to 128MB. As the CPU is only a 186 it doesn't support extended memory so most DOS applications will only be able to see 640k no matter what. It's really only useful as storage or for EMS if something supports that.
1
"big.LITTLE" is ARM, not Qualcomm.
1
@GamersNexus Obviously it's a dead CPU but was it dead before the hammering? Reminds me of when I tried to overclock a Pentium 4 by breaking off one of the BSEL pins but ended up breaking a more important pin instead and trashing the processor.
1
The thing is 4800U sounds like a name I'd expect Intel to use. The name fits right in with their earlier 4200U and 4300U Ultrabook chips from a few years ago.
1
Intel can easily show themselves above AMD on benchmarks. They just need to take a pocketknife and carve themselves on the backrest of the bench and AMD on the seat of the bench. Do I hear crickets?
1
The site JonnyGURU has some nice technical reviews of PSUs. They've reviewed a number of these junk PSUs (search for 'gutless wonder') and many completely die at half the advertised power. I got a used case which came with an LP-8860 PSU which they reviewed. Mine came pre-dead and the quality on it is laughable.
1
@ThoArtOne I wouldn't say all of them are low quality but it does seem hard to find a good smaller-capacity PSU at retail. Higher-end manufacturers tend to focus on the higher-wattage ones because that's where the money is. If you want a good small-capacity PSU your best bet is probably to pull one from a business-grade PC. The problem is these are usually more like 250w, probably only good for IGP systems.
1
@00000005547 Most server applications rely on random disk access IOPS instead of sequential speed. A single HDD may give 200 MB/sec sequentially but on random access you'll get 100 IOPS on a 7200rpm drive if you're lucky, equating to under 10 MB/sec. The cheapest consumer SSD will give you 10 times this, high end U.2 ones 1000x. Per IOPS SSDs end up being way cheaper than HDDs, not to mention the space and power savings. For reliability, why do you say HDDs always give warning? Sure they do sometimes, but the same can be said of SSDs. SSDs have a very clear wearout indicator which gives you a great idea of drive aging. You'll still have sudden controller failures but HDDs are the same. If you look at a server vendor's website SSDs are in most of the preconfigured options, and not without reason.
1
Believe it or not the G200 is still used today in modern servers.
1
And then AMD just names their CPUs based on "equivalent MHz".
1