dom0

90 ms·

841.
by dom0 10y ago
> The software industry requires a legislative bitch slap like the auto industry received. These rules would wreak havoc on the industry but if you ask me for the better. When did that ever happen?
842.
by dom0 10y ago
> If the intention was just to speed up hand-tuned library functions, it would be much easier to ask the library-hand-tuners to figure out instruction dependencies ahead of time (instead of doing re-ordering) and to provide a lot of regi
843.
by dom0 10y ago
Intel pricing and performance increases since AMD went away (but it seems to be coming back now, so that's good - and indeed, Intel already cut prices in some segments).
844.
by dom0 10y ago
> So during those 5 instructions, theoretically it could be doing some of the next hundred. Not if the next iteration of the loop not only branches on the previous result (okay, that one's easy on the predictors), but also depends o
845.
by dom0 10y ago
The µop cache (trace cache went out the window) doesn't help with fetching instructions AIUI, but accelerates decoding them. So the 4 insns/cycle fetch bandwidth is not changed by that.
846.
by dom0 10y ago
I don't think it's even possible any more. Haswell and later are 8-wide designs, but can only fetch+decode and retire 4 instructions per cycle (peak). So one would need ops that decode to 2+ execution-unit independent µops. Line
847.
by dom0 10y ago
There is also a latency penalty for accessing PCIe devices that are attached to a foreign root complex. This is exacerbated if the QPI links are already busy.
848.
by dom0 10y ago
Making decisions and thinking what's good for the ecosystem are two different things. I can still buy nVidia for my company, and think and say at the same time that it's bad. Of course, while it's the cheap consumer garbage ^
849.
by dom0 10y ago
In the long run every monopoly hurts badly. Puzzling that people think it's good for GPGU, with the x86 fiasco next door.
850.
by dom0 10y ago
> Is unoptimized code really that bad? Most application code is far worse and stalls (sadly - almost literally) every few instructions for L2 / memory. > (Code is small relative to data, most of it should be in cache most of the
851.
by dom0 10y ago
> lightweight site with a good design and readability: http://www.zeldman.com/ >1 MB is not "lightweight".
852.
by dom0 10y ago
Hm, maybe give Borg a shot. For initial backups it's not exactly the fastest thing, though much better after that.
853.
by dom0 10y ago
Uhm, what exactly is crashplan doing with all that memory?
854.
by dom0 10y ago
I use quite a bit of Supermicro hardware and have to conclude that it's just good stuff (tm). Recently had to work with an Intel server board, ick ... no comparison at all.
855.
by dom0 10y ago
The main issue with used server hardware that is a couple generations old is that they use a lot of power. E.g. it's no problem to buy dual or quad port Intel server adapters for like 20 Euros, but these draw 6 / 10 W, while newer
856.
by dom0 10y ago
Volksempfänger would be damn classy propaganda. But Trump is not classy, never was, and never will be.
857.
by dom0 10y ago
> Elbrus, VLIW, Transmeta, Intel It's interesting how much people Intel soaked up in the early 2000s after the RISCollapse and VLIW-gone. Sometimes I wonder where exactly Intel got all the money from for doing that and pushing throu
858.
by dom0 10y ago
Can you elaborate? I only removed the qualifier ("written by non-native / lazy / busy devs"), since almost any text that has not been copy-edited can be improved by doing just that, which requires little or no domain kno
859.
by dom0 10y ago
CodeXL can do that (at least on Windows) I think VTune can do OpenCL analyses, I don't know whether it has any graphics debugging capability (quite a lot of features are Windows-only or annoying to set up on Linux [because who needs a
860.
by dom0 10y ago
Humans and many other animals have been eating meat for billions of man-years (this is a software site after all), I kinda doubt "actually quite dangerous" is a fitting description. Bungee-jumping without a rope is "actually
861.
by dom0 10y ago
Eh, you can already debug kernels on a line-stepping level. What exactly do you need more?
862.
by dom0 10y ago
Well the Concorde was twice as fast (record: 2 hours and 52 minutes), but it's decommissioned now.
863.
by dom0 10y ago
> * documentation written by non-native speakers (and lazy/busy devs in general) needs a lot of editing and corrections. ftfy: > * documentation needs a lot of editing and corrections. Good docs can be as much work or more as the
864.
by dom0 10y ago
> To be honest, a PR from someone who doesn't use your software, who is unfamiliar with its structure and only wants her PR merged so that she can say 'I am an open source contributor' is a hassle. This is a reason why man
865.
by dom0 10y ago
As usual, no-execute bits in the page table is something that trickles down. I think in x86 land AMD introduced it with AMD64. The RISCies (Sparc, Power, Alpha, hppa) had this earlier.
866.
by dom0 10y ago
I don't know much about ZFS' deduplication, just heard that it requires a lot of memory, in a "hard minimum amount" way, to do it. This suggests, to me, that at least one design element of their deduplication engine is p
867.
by dom0 10y ago
> I always got the impression (since I started using it in Prod in the late 90s) that the MySQL team had never worked on an RDBMS before, even as users, and didn't really understand it. IIRC that's actually correct.
868.
Germany’s right attacked the media. Now Germans trust the press more than ever(washingtonpost.com)
1 points by dom0 10y ago|0 comments
869.
by dom0 10y ago
Next up from The Author: 5 Reasons Why We Switched from MySQL to Postgres Why We Switched from Postgres to Oracle 11 Reasons For Switching From Oracle To KDB 7 Reasons Why We Switched from Go to Fortran-77 Grammar errors are provided for fr
870.
by dom0 10y ago
> Not implementing TLS because you're worried about attack surface is a bad idea. [citation needed]

More ›