Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
baruch
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
31.
▲
Cache Awareness in DPDK Mempool
(dpdk.org)
2 points
by
baruch
2y ago
|
1 comments
32.
▲
by
baruch
3y ago
Just the other day I saw a local company get investment to be cybersecurity for LLMs. It's here already.
33.
▲
by
baruch
4y ago
Calculating the divisor values is also expensive, this works when you do this work once and then do the efficient divides multiple times.
34.
▲
by
baruch
4y ago
I know that this idea nugget comes up time and time again when discussing drives, but I'm working on large storage systems for over 10 years now and dealing with the drives and their reliability and have yet to see such an occurrence.
35.
▲
by
baruch
4y ago
The main reason I took my first management position was after I remembered the pain of refusing a previous offer and getting an awful manager to lead the team. Though I do not think I'm very good at managing myself. Though I might be a
36.
▲
by
baruch
4y ago
Any Intel server comes with an ARM chip (at least one that I know of), the BMC.
37.
▲
by
baruch
4y ago
One you send humans that you can't sterilize before flight you will inevitably introduce bacteria into the environment anyway, and you will not be preselecting them either.
38.
▲
by
baruch
4y ago
That was Shunra, I worked there for a brief time. It was sold to HP at some point and I don't know what happened to the product afterwards.
39.
▲
by
baruch
4y ago
Normally the endpoint stack you need is more limited initially and you don't have to have all the baggage. The biggest advantage you get at the end is zero copy network and if what you do involves tons of reading and writing to the net
40.
▲
by
baruch
4y ago
DPDK has an XDP driver too so you can use that and also work with network cards that do not support XDP or the zero copy variant for more flexibility. Though DPDK can be a pain in and of itself, it is fairly opinionated and I don't li
41.
▲
by
baruch
4y ago
I have no beef in this, I'm not living in the US and not Indian or south Asian. The thing that is the issue, again from the little I read, is that such discrimination is brought by other Indians on the Dalit and it's most defyingl
42.
▲
by
baruch
4y ago
From what I read so far the discrimination is not wasps but rather by other Indians. So no one is saying you are discriminating one care or another but that such discrimination exists even if you don't see it.
43.
▲
by
baruch
4y ago
There was also BiModem or something like that for the higher speeds too. With restart recovery and larger block sizes.
44.
▲
by
baruch
4y ago
It depends on your commander, you can buff your commander and tell him he is suggesting something stupid, though if he pushes it you have no real option but to do it. Though if the specific commander was all too happy to assert his rank you
45.
▲
by
baruch
5y ago
The generated function also has a subtract zero from rsp which could just be plain eliminated. Though it's probably rarer to have this in general code, it means there are no local variables. Looks like there are still some low hanging
46.
▲
by
baruch
5y ago
I use git-cola mostly for the partial diff commit feature. A text interface tool will be nice to try though.
47.
▲
by
baruch
5y ago
Part of the pain with kernel bypass is memory management. For real performance gains you need zero copy and for that you need to work inside the packets. Perfectly doable for a green field project not so east I think for a large existing sy
48.
▲
by
baruch
5y ago
At a higher level, considering that all protocols are serial these days, how does a multi-lane protocol (PCIe x4 or x16, Ethernet x4) use the multiple lanes, is it multiple independent requests going on each lane?
49.
▲
by
baruch
5y ago
Is it the SerDes component that also does the word encoding (8b10, etc., forgot the proper terminology for this) ? I work on the Hardware/Software interface quite a bit (more in the past than now) and this is always fascinating, I'
50.
▲
by
baruch
5y ago
The SATA protocol itself is rather outdated, for example, when using NCQ and if a single command gets an error (say, medium error) all the other pending commands are aborted and you need to resubmit them. That's quite a pain and a perf
51.
▲
by
baruch
5y ago
Presumably there is a cost to carrying the DNA, if it gets removed by chance and the cost was high enough the useless part will disappear. That however hinges on the cost being high enough, if the cost is negligible then there is no real pr
52.
▲
by
baruch
5y ago
Most drives these days are TLC and and consumers are more likely on QLC. I havent even seen an MLC drive for a while now.
53.
▲
by
baruch
5y ago
Do you sources for the 1 million writes(erase) on MLC NAND? I work with enterprise SSDs and that number is well above anything I've seen or heard, even from the manufacturers.
54.
▲
by
baruch
5y ago
In PCIe gen 3 it is already possible to saturate the PCI lanes with NAND. Even in gen 4 you get 6 GB/s. You gain some in latency but you really need to invest in the software to eke out that latency gain. 3DXpoint really shines in pers
55.
▲
by
baruch
5y ago
I work on the WekaFS (proprietary) that can definitely handle the relevant workloads and is in use in fairly large installations.
56.
▲
by
baruch
5y ago
Networking actually has tons of interesting and complex math. Congestion control is a rabbit hole of math and control theory.
57.
▲
by
baruch
6y ago
I work on an SSD storage system with multiple SSDs in any server and 100Gbps NICs are the low end for us. We see significant use for 200 Gbps.
58.
▲
by
baruch
6y ago
Wouldn't this mean that you need to memcpy around? Maybe async-io (io_uring?) help here by doing zero-copy writes directly from the source. I don't know how much you need to mangle code (GOT/PLT, offsets) and how much it is a
59.
▲
by
baruch
6y ago
One way to speed up IO-bound code is to have multiple threads-of-execution (threads or async-io) in order to have multiple requests in the air. Specifically SSDs benefit a lot from multiple requests in parallel as that utilizes their intern
60.
▲
by
baruch
6y ago
BetterPlace was a rent seeker that lobbied to limit ability to use other service providers including for example the home charging station. They lobbied to have the home chargers regulated so that only they could provide it. They also spent
More ›