ASRock Debuts AM4 Motherboards With 5 Gigabit LAN

[bit_user]

This is pathetic: I'm waiting for more than a decade to see 10 Gbit into the mainstream MB and they timidly start with 2.5 and 5 Gbit, 15 years after the 10 Gbit standard was approved...

You haven't listened to a 10 Gigabit switch, I take it? They're not quiet. Or cheap.

For this reason, I welcome 5 & 2.5 Gigabit. I guess stepping stones were needed to reach 10. I think it'll actually happen sooner, this way. Too bad they weren't introduced years ago.

I kinda wish they'd just gone with 4 Gigabit, instead of 2.5 and 5. If 2.5 and 5 were at least introduced a few years apart, I could understand. But together might overestimate how much appetite the market has.

 

[bit_user]

Even today, most people continue to use hard disk drives for bulk storage, and their performance hasn't improved all that much, only being around twice as fast as they were then. Under a best case scenario, a modern disk drive might potentially allow for slightly higher bulk transfer speeds than what a 1 Gbit network connection can handle, but it's not going to even come close to what a 2.5 Gbit network is capable of.

This is nonsense.

http://www.storagereview.com/wd_black_6tb_hdd_review

I purposely didn't go for a 10 TB model, since that's legitimately cutting edge.

So, this comes back to why you haven't seen mainstream motherboards supporting 10 Gbit Ethernet. Few people have had a bulk storage solution fast enough to take advantage of more than what regular Gigabit Ethernet can offer, so it would have largely been a waste putting that feature on boards. Putting a costly feature that very few people will have need for on a motherboard for would just raise the price of the board for everyone.

Heard of NVMe? NVMe SSDs can often do north of 1 GB/sec.

Therefore, it's not unreasonable to put 10GBaseT on boards with at least one M.2 NVMe slot. Of course, NVMe drives are available as ordinary PCIe cards (I have one), but if someone buys a board with a M.2 NVMe slot, then there's a decent likelihood it's gonna get populated.

 

[dgingeri]

Even today, most people continue to use hard disk drives for bulk storage, and their performance hasn't improved all that much, only being around twice as fast as they were then. Under a best case scenario, a modern disk drive might potentially allow for slightly higher bulk transfer speeds than what a 1 Gbit network connection can handle, but it's not going to even come close to what a 2.5 Gbit network is capable of.

This is nonsense.

http://www.storagereview.com/wd_black_6tb_hdd_review

I purposely didn't go for a 10 TB model, since that's legitimately cutting edge.

So, this comes back to why you haven't seen mainstream motherboards supporting 10 Gbit Ethernet. Few people have had a bulk storage solution fast enough to take advantage of more than what regular Gigabit Ethernet can offer, so it would have largely been a waste putting that feature on boards. Putting a costly feature that very few people will have need for on a motherboard for would just raise the price of the board for everyone.

Heard of NVMe? NVMe SSDs can often do north of 1 GB/sec.

Therefore, it's not unreasonable to put 10GBaseT on boards with at least one M.2 NVMe slot. Of course, NVMe drives are available as ordinary PCIe cards (I have one), but if someone buys a board with a M.2 NVMe slot, then there's a decent likelihood it's gonna get populated.

My storage server has 6 4TB drives in a RAID 10 which can push north of 400MB/s. It could easily saturate a 2.5Gb/s ethernet connection, and come close to saturating a 5Gb/s link. I agree with you that the increase in bandwidth is needed, preferably up to 10Gb/s. We need to get 10Gbase-T reduced in price.

 

[bit_user]

My storage server has 6 4TB drives in a RAID 10 which can push north of 400MB/s.

I can do about 350 MB/sec with the software RAID 6 of 5x 1 TB drives that I built in 2011. But I purposely didn't pull that one out because @cryoburner was trying to make a point about mass market needs. SSDs are even faster, and now truly mass market.

IMO, @cryoburner needs to wake up and smell the SSDs!

 

[JonDol]

This is pathetic: I'm waiting for more than a decade to see 10 Gbit into the mainstream MB and they timidly start with 2.5 and 5 Gbit, 15 years after the 10 Gbit standard was approved...

You haven't listened to a 10 Gigabit switch, I take it? They're not quiet. Or cheap.

For this reason, I welcome 5 & 2.5 Gigabit. I guess stepping stones were needed to reach 10. I think it'll actually happen sooner, this way. Too bad they weren't introduced years ago.

I kinda wish they'd just gone with 4 Gigabit, instead of 2.5 and 5. If 2.5 and 5 were at least introduced a few years apart, I could understand. But together might overestimate how much appetite the market has.

That was not the point. I'm aware the current 10G switches are actively cooled but that won't make more noise than my PC or my NAS. And I neither asked them to be quiet nor cheap. Just integrated into the consumer (not prosumer) products i.e. easily buyable (without going to specialized resellers which won't sell them to you because you are not a company).

 

[bit_user]

I'm aware the current 10G switches are actively cooled but that won't make more noise than my PC or my NAS.

Perhaps not, if your PC and NAS are 1U. The problem with actively-cooled switches is that the fans must spin at high RPMs, to move sufficient air with such small blades. As new switch chipsets & transceivers are manufactured on smaller lithographies, they'll hopefully become more efficient, require less cooling & commensurately fewer RPMs.

And I neither asked them to be quiet nor cheap.

Perhaps not, but these are issues for many would-be buyers. And volume is needed to drive down prices and increase market participation among manufacturers and vendors.