General Computer Thread

[Marc]

Harddrives are pretty much nearing the limit of what you can cram on a platter, they're going for heat assisted writing and microwave assisted writing and a few other techniques like that but there's only so much you can do, guess we might see 5'25" drives again.

I'm not sure that will happen.

With density a very important factor these days 2 x 20TB (in early OEM testing) drives in 3.5" factor would be seen as a better option over say a 35TB drive in 5.25" form because the case would have to be so much larger to fit in the physical larger drive.

plus it's better for fault tolerance.

And some of the new methods to increase capacity need to just go away and die. a lot of people were pissed off to find they'd bought drives that use SMR without realising it.

They could always make 5,25" drives available again via a new technology that allows to put more on a disk, couldn't they? Then again, the memory stick sort of has taken on that role.

not really.

They still don't have the storage capacity of spinning rust and neither do they have anywhere near the speed. Reaching the point where if you need a large capacity USB drive the best option is a SSD in an external case.

 

[Kamen Rider Blade]

I'm not sure that will happen.

I don't think that will happen, just because the Form Factor fell out of favor for many reasons and 3.5" HDD's have been the dominant mass storage form factor for so long, any change will be unlikely due to the low profit margins on HDD's.

3.5" HDD still have a long way to go with HAMR / MAMR technologies and Seagate with it's Dual Actuator tech is exciting for finally bringing more R/W throughput finally in a big way. Now if Seagate can bring that Dual Actuator tech to the consumer end and allow us to finally get faster than SATA III speeds, I'll be a happy camper.

They say the only penalty for Dual or Multi Actuator tech is losing 1 platter of density, which is fine by me, if you can shove that much more speed into it with Multi Actuator, that would be AMAZING. Imagine Reading/Writing at ≥ 1 GiB/s

That would be AMAZING since they can finally port over SAS-3 with 12 Gbps and use newer encoding technologies to maximize the bandwidth.

For a Hypothetical SATA 3.5 standard using 128/130b encoding, you can get 750 MiB/s Theoretical -> ~738.462 MiB/s actual throughput and do it with Full Duplex capabilities.

For a Hypothetical SATA 4.0 standard using 128/130b encoding, you can get 12000 Mb/s Theoretical -> ~1476.923 MiB/s ~= 1.442 GiB/s actual throughput and do it with Full Duplex capabilities.

That would be awesome if we consumers can get such nice things. =D

 

[Gingerbread Demon]

Is there a theoretical limit to how fast CPUs can go at the moment?

 

[Kamen Rider Blade]

Is there a theoretical limit to how fast CPUs can go at the moment?

Not to my understanding, but I could be wrong.

Right now we have a Thermal Barrier, it gets harder to cool the CPU's due to the density of the transistors and cooling so much heat in such a tiny area.

This past year at HotChips 2020, this was the first major step to Silicon Photonics.

https://www.anandtech.com/show/16010/hot-chips-2020-live-blog-silicon-photonics-for-ai-600pm-pt

The amount of energy needed along with transmission can be faster if we move to transmission via light instead of electricity for certain things, but the R&D behind it has been in the very beginning.

This year there were some serious break throughs that will get more eyeballs onto it.

 

[Gingerbread Demon]

Oh god can you imagine AMD optical CPU, and the marketing slogan
"now there are no limits"

 

[Kamen Rider Blade]

Oh god can you imagine AMD optical CPU, and the marketing slogan
"now there are no limits"

Imagine once we discover how to manipulate SubSpace and break the FTL barrier with processing.

Because that's how the TNG Tech manuals stated Trek Computers got beyond the FTL barrier, there was a mini SubSpace field imbued on the processors to go Super Luminal Computing, ergo the signals weren't limited by the Speed of Light!

=D

 

[Marc]

For a Hypothetical SATA 3.5 standard using 128/130b encoding, you can get 750 MiB/s Theoretical -> ~738.462 MiB/s actual throughput and do it with Full Duplex capabilities.

For a Hypothetical SATA 4.0 standard using 128/130b encoding, you can get 12000 Mb/s Theoretical -> ~1476.923 MiB/s ~= 1.442 GiB/s actual throughput and do it with Full Duplex capabilities.

That would be awesome if we consumers can get such nice things. =D

I'm not sure it would be cost and time effective to bring that to the consumer market.

New technologies are coming along that will be bring down the prices of SSD at the same time as increasing capacity combined with M.2 NVMe interfaces that tie directly into the PCIe bus would be enough to satisfy 95% of consumers.

AMD has PCIe 4.0 support in higher end motherboards and that will filter down. Not sure where Intel is with the new standard but probably not that far away.

Tie in with a M.2 NMVe supporting PCIe4 and watch the thing pump through upto 5000MB/s (SAMSUNG Evo 980).

I don't think we'll see SATA go full duplex/dual port as a) not worth the effort b) it already exists with SAS and vendors wouldn't want to sabotage that market (a long standing attitude that's probably done a lot to hold things back)

 

[Kamen Rider Blade]

I don't think we'll see SATA go full duplex/dual port as a) not worth the effort b) it already exists with SAS and vendors wouldn't want to sabotage that market (a long standing attitude that's probably done a lot to hold things back)

If the SAS community is worried about getting it's territory encroached upon, there's no reason "Dual Port" wouldn't be a thing on consumer drives, so they would have an exclusive feature.

As far as I can tell, the "Full Duplex" / "Half Duplex" was a artificial limitation imposed on SATA by SAS.

As far as M.2 NVME, that interface wasn't really designed to be "Hot Swapped". the port's tested insertion durability isn't particularly high since it was never designed to be "Hot Swapped" in and out.

U.2 hasn't really taken off and is a more expensive interface on mass production level.

You still need a solution for the budget end that doesn't go full M.2/U.2 speeds.

Kind of like how we have Ethernet 1Gbps & 10 Gbps, but they still made a newer but slower 5 Gbps and 2.5 Gbps standard.

 

[Marc]

If the SAS community is worried about getting it's territory encroached upon, there's no reason "Dual Port" wouldn't be a thing on consumer drives, so they would have an exclusive feature.

As far as I can tell, the "Full Duplex" / "Half Duplex" was a artificial limitation imposed on SATA by SAS.

As far as M.2 NVME, that interface wasn't really designed to be "Hot Swapped". the port's tested insertion durability isn't particularly high since it was never designed to be "Hot Swapped" in and out.

Hell the drives are screwed in which is the antithesis of hotswap Though I was taking your point about consumer space and running with it and hotswap is more an business that consumer market thing (unless you're running a homelab in which case some of them put business setups to shame

Had a motherboard with U.2 port on it, never used the thing (think it was on my Athlon board). does seem to be still be around in the enterprise market which they have even more choice for drive interfaces (all the above plus EDSFF)

Kind of like how we have Ethernet 1Gbps & 10 Gbps, but they still made a newer but slower 5 Gbps and 2.5 Gbps standard.

[/quote]

Depends on whether you're talking 10Gbps over UTP or using SFP+ (DAC/Fibre)

Over UTP was pretty much DOA, chipsets were very expensive and by all accounts those things were power hungry and ran very hot. 2.5 and 5Gbps was way of meeting the demand but avoiding the pitfalls of the original 10Gbps

 

[Kamen Rider Blade]

Depends on whether you're talking 10Gbps over UTP or using SFP+ (DAC/Fibre)

Over UTP was pretty much DOA, chipsets were very expensive and by all accounts those things were power hungry and ran very hot. 2.5 and 5Gbps was way of meeting the demand but avoiding the pitfalls of the original 10Gbps

2.5 & 5 Gbps are over existing CAT-5 / CAT-6 cables.

 

[Marc]

2.5 & 5 Gbps are over existing CAT-5 / CAT-6 cables.

I know.

That they can do it without the issues of 10Gbit over UTP makes it more viable

Though 2.5 seems to the only one making headway at the moment (cards and switches available) as well as integrated into some of the higher end motherboards ($300+)

 

[Santaman]

I started with 10mbit, coax cables and terminators.. lot of things changed over the years..

 

[Marc]

I started with 10mbit, coax cables and terminators.. lot of things changed over the years..

unplug the wrong cable and the whole thing comes crashing down

There were worst things though.

At one point I was dealing with the local branch of an airline and they used an IBM AS/400 with token ring. Disconnecting from the emulator cards in the PCs was okay, breaking the loop ring was another as it required a power cycle on the 400 and that took an age.

Never made a UTP cable but did once with coax and never again (might have gone better if the company tools were a bit better cos it's easy in theory).

Was out at site recently and found they still had a hub (no not a switch this definitely a hub with a coax port on the back) and first time I'd seen one in over 15 years.

 

[Santaman]

Found parts cheap and replaced the Sempron 145 with a Pentium Gold 6400.. yeah quite an upgrade for the old beast, even moreso because the casing is from a Medion machine from the late '90's which originally housed a Pentium 4 2.66Ghz.

 

[Dryson]

Does anyone know what would cause the tethering between a smartphone and PC to suddenly not function that would require a replacement smartphone?

I purchased a brand new cord this morning and still didn't get the tethering option to appear. The smartphone is still able to take a charge though.

 

[Kraig]

My old Moto Z droid won't tether until you select data transfer on the phone, then you have to shut the phone screen off. Sometimes the cord can be the problem.

 

[Santaman]

Maybe reboot the phone since it is essentially a small computer, maybe that works.

 

[Gingerbread Demon]

Check the cord if all else fails and try a different cord. It can be as simple as that.

Also trying to install Linux Mint on an Acer A315 this machine has a regular 2.5 inch Sata and also a hidden m.2 slot which is the drive I am currently using on this machine. Took that out to use the regular drive bay for my install, keep getting "secure boot fail" when the machine reboots with picture of a giant padlock on the screen. Have tried installing with bios set to legacy but can't turn off the actual secure boot for the SSD even in the regular drive bay. Have tried clearing all the passwords for the drive and clearing the TPM too.

Google wasn't helpful as this machine wasn't a very widely used machine and not popular because it's the version with an E2 chip and not a regular AMD or Intel chip.

 

[Gingerbread Demon]

How often do you guys and gals open your PC and give it a good clean?

 

[Deks]

Is there a theoretical limit to how fast CPUs can go at the moment?

You can hit a limit with specific materials such as silicon (at which point you'd need to transition to new materials).
Arguably, we should have started using different materials in computer chip designs such as synthetic diamonds, carbon nanotubes and graphene a while ago (at least where appropriate in the computer chip) to make proverbial hybrids... but markets don't work like this.

They tend to focus on 'cost efficiency' and existing production/manufacturing (and what's worse, many prefer to wait so they can produce new materials via existing methodology - which is mainly driven by money not scientific or technical breakthroughs - otherwise, nothing is really stopping large businesses to say harvest existing production facilities for their raw materials, decompose them into base elements and then use them to build new/more suitable facilities - but again, they don't think like this as 'cost' is again a limiting factor to them)... so if transitioning to say new materials would require redesign of production facilities, it will only be done if its cost-effective (monetarily affordable) and profitable (otherwise it COULD have been done from a resource and scientific/technical point of view a while ago).

Hybrid computer chips, or complete discarding of silicon as a material and how computer chips work might be next.
For example, graphene doesn't technically have an 'off' switch so you need to make one (which was designed a LONG time ago actually), but to be fair how we LOOK at computer chip operations also might need to change.
If we keep trying to apply old methodologies to new materials as opposed to trying to make chips by making use of new material natural properties as they are... it would take AGES to do stuff (Well, not anymore, because you can actually put a supercomputer with an AI or adaptive algorithm to do the research and it can be done in a fraction of a time that it would take whole teams of specialized people to do that).

In regards to HDD's reaching their limits... I agree, and to be fair, I actually HATE how they keep dragging out the platter HDD's so much.
IBM already developed ideas for holographic storage in the 90ies... but if you want something 'less exotic'... then SSD's are more likely to take over for HDD's when it comes to storage density and reliability... but cost is an issue here because a 4TB SSD for example can cost as much as a 16TB HDD... which is absurd (seeing how prices of SSD's should have actually beaten HDD's by now).