|
dont be mean to me posted:When was the last time browsing the Web was light duty? loving this. Websites are turning into steaming piles of poo poo that eat 1000 gigs of ram and 37 cores just by opening them. If don't want to close your browser every time you want to play a game, buy something with 8 threads.
|
#
?
Jun 11, 2017 13:56
|
|
|
|
| # ? Apr 18, 2024 01:25 |
|
|
Measly Twerp posted:Since it might be interesting to others I'm going to benchmark my system before and after the upgrade. Cinebench, both single and multithread.
|
|
#
?
Jun 11, 2017 14:22
|
|
|
Saukkis posted:There's also the question of if you are upgrading from i3-6100 would something like i5-7600K or i7-7700K be a better option than an R7 system. You could skip buying a new motherboard. when guys have been posting a bunch of prices in euros and hungarian ft it's also good to forget assumptions like these. between regional price differences and VAT exaggerating the spread between two prices, it's entirely possible that a 7700k *does* cost more than a ryzen + new mobo for him.
|
|
#
?
Jun 11, 2017 14:22
|
|
|
Mr.Radar posted:Cinebench, both single and multithread. It is unfortunately unavailable on Linux. Klyith posted:when guys have been posting a bunch of prices in euros and hungarian ft it's also good to forget assumptions like these. between regional price differences and VAT exaggerating the spread between two prices, it's entirely possible that a 7700k *does* cost more than a ryzen + new mobo for him. While it does cost more than a 1700, it doesn't cost more than that combined with a motherboard. It's just unappealing due to having half the cores. vvv Forgot about a cooler... Anarchist Mae fucked around with this message at 14:41 on Jun 11, 2017 |
|
#
?
Jun 11, 2017 14:35
|
|
|
Klyith posted:when guys have been posting a bunch of prices in euros and hungarian ft it's also good to forget assumptions like these. between regional price differences and VAT exaggerating the spread between two prices, it's entirely possible that a 7700k *does* cost more than a ryzen + new mobo for him. Anime Schoolgirl fucked around with this message at 14:40 on Jun 11, 2017 |
|
#
?
Jun 11, 2017 14:37
|
|
|
Measly Twerp posted:It is unfortunately unavailable on Linux. According to the WINE DB it runs perfectly with WINE, if you're not adverse to installing WINE.
|
|
#
?
Jun 11, 2017 14:52
|
|
|
Truga posted:loving this. Websites are turning into steaming piles of poo poo that eat 1000 gigs of ram and 37 cores just by opening them. If don't want to close your browser every time you want to play a game, buy something with 8 threads. Yeah totally. That's why it's always funny when you see people asking for suggestions for a laptop for their mom or something and they say "oh she's only going to be surfing Facebook!". That'll loving eat a huge chunk out of a decent CPU these days.
|
|
#
?
Jun 11, 2017 16:18
|
|
|
The people that browse the web on 6100s for light duty are better served using phones or tablets they likely already own. System integrators that sell 6100 systems frequently pair it with 4gb and a spindle drive. 2c4t for light office work fine that is a use case but you'd also need a ssd and at at the very least 8gb of ram. i3s certainly have a place given their value but let's not kid ourselves that it dumpsters a 1700.
|
|
#
?
Jun 11, 2017 18:57
|
|
Risky Bisquick posted:The people that browse the web on 6100s for light duty are better served using phones or tablets they likely already own. System integrators that sell 6100 systems frequently pair it with 4gb and a spindle drive. 2c4t for light office work fine that is a use case but you'd also need a ssd and at at the very least 8gb of ram. i3s certainly have a place given their value but let's not kid ourselves that it dumpsters a 1700. Ehhh, not really, with the Pentiums having 2c/4t now it's far better to just go with a G4560 for far less money.
|
|
|
#
?
Jun 11, 2017 19:09
|
|
|
Yea I don't really see a place for the i3 anymore now that pentiums have HT and Ryzen 1400/1500 exist for sub $200 with the 1600 costing not much more after that.
|
|
#
?
Jun 11, 2017 19:28
|
|
|
Now that I'm back from vacation, I finally had the time to sit down and tinker with the B350 to figure out why my DDR4-2400 was only running at 2133. The solution was to disable A-XMP entirely, manually input the timings and the voltage, and then CPU-ID finally correctly showed that both sticks were running at 2400. I heard that flashing MSI's latest beta BIOS also helps, but I'd rather not take a chance now that it's all sitting pretty. Hooray for an extra ~5-6% performance increase! fishmech posted:Yeah totally. That's why it's always funny when you see people asking for suggestions for a laptop for their mom or something and they say "oh she's only going to be surfing Facebook!". That'll loving eat a huge chunk out of a decent CPU these days. Facebook is the most bloated, lovely, terribly programmed social media website. I have no idea what fuckery the webdevs are doing to make it run so sluggish, even on good hardware. Even trying to just load the newsfeed occasionally gives my PC a hernia, so I just don't even bother with the website anymore.
|
|
#
?
Jun 11, 2017 20:06
|
|
|
It's a gently caress load of Javascript and libraries.
|
|
#
?
Jun 11, 2017 20:21
|
|
|
Facebook has really bad development practices in general, like when their Android app had too many methods so they decided to invasively modify the Android VM on app startup
|
|
#
?
Jun 12, 2017 01:21
|
|
|
It honestly really depends on the game, in older games or those that load 2 or so cores, the i3 6100 might compete with an overclocked Ryzen, but there are several instances where an i3-7350K @ 4.8Ghz loses (sometimes really badly) and thus an i3 6100 would have no chance. https://www.youtube.com/watch?v=O5cqOtWz5sU Clockspeed isn't everything and it'd be tough to recommend an i3 or i5 right now compared to the 1500X or 1600. The extra cores and cache really seem to carry Ryzen.
|
|
#
?
Jun 12, 2017 05:30
|
|
|
 It's finally here! So far I've not configured my RAM, or thought about overclocking. Also, GRINDCORE MEGGIDO asked a mod for a 'funny probation' and got probated for 11 years...
|
|
#
?
Jun 13, 2017 12:38
|
|
|
Measly Twerp posted:Also, GRINDCORE MEGGIDO asked a mod for a 'funny probation' and got probated for 11 years... I felt bulldozered, but now I am Ryzen. E-(I got one. I knew Palpek would deliver). GRINDCORE MEGGIDO fucked around with this message at 01:05 on Jun 14, 2017 |
|
#
?
Jun 14, 2017 00:29
|
|
|
Here are my benchmark results: http://openbenchmarking.org/result/...-STOA-170613155
|
|
#
?
Jun 14, 2017 01:24
|
|
|
Linux compile owns.
|
|
#
?
Jun 14, 2017 01:29
|
|
|
I'm guessing SciMark uses AVX?
|
|
#
?
Jun 14, 2017 03:22
|
|
|
Turdsdown Tom posted:
the naivety of someone who has never seen tumblr
|
|
#
?
Jun 14, 2017 04:30
|
|
|
how cleanly does (python) scientific code parallelize? i'm tempted to throw money at a ryzen r7 1700 as an upgrade over my current i5 4590; would I get close to 2x performance moving from an i5 4 core to ryzen 8 core for hobby machine learning workloads (e.g., sklearn, xboost)?
|
|
#
?
Jun 14, 2017 06:56
|
|
|
RME posted:the naivety of someone who has never seen tumblr I watched a video gallery in tumblr bring a pair of E5-4620 V4's to 100% cpu utilization.
|
|
#
?
Jun 14, 2017 07:16
|
|
|
shrike82 posted:how cleanly does (python) scientific code parallelize? i'm tempted to throw money at a ryzen r7 1700 as an upgrade over my current i5 4590; would I get close to 2x performance moving from an i5 4 core to ryzen 8 core for hobby machine learning workloads (e.g., sklearn, xboost)? If you are tying down into numpy/scipy (numexpr is another one that came up on a StackOverflow response), i.e. something that hooks native C code and runs operations in parallel on a matrix, then it's pretty good. Raw Python code, it's garbage. Interpreted Python does not thread due to fundamental architectural flaws that the BDFL will not change. You can have "threads" that take advantage of explicit yields or blocked I/O, but you will never have two threads actually executing at the same time due to the GIL, so the only way you can write truly parallel Python code is process-level parallelism. Or by going to another interpreter like Jython but you don't have any guarantee that any given library is written around an assumption of concurrency, so this is likely to crash and burn horribly. Fixing this would basically involve going back and editing every Python package ever written, and adding thread-safety to it. Now, Python 3 actually requires everyone to go back and manually edit every Python 2 program for compatibility anyway... which has already taken almost a decade and produced a deep fracture in the Python community... but this particular breaking change was just one breaking change too many, according to GVR... Paul MaudDib fucked around with this message at 07:27 on Jun 14, 2017 |
|
#
?
Jun 14, 2017 07:22
|
|
|
yah i'm aware of the GIL - the code and frameworks I use spin up new processes. my question is whether i'll get a straight 2x performance boost moving from an i5 4 core to ryzen 8 cores or do the architectural differences blunt that substantially.
|
|
#
?
Jun 14, 2017 07:28
|
|
|
shrike82 posted:yah i'm aware of the GIL - the code and frameworks I use spin up new processes. Matrix math tends to parallelize real well thanks to time-tested primitives like BLAS/LAPACK (which Numpy hooks as native C). I'm guessing if you're getting good scaling (4x speedup) on a 4-core already then you would get 8x out of Ryzen, yes.
|
|
#
?
Jun 14, 2017 07:29
|
|
|
nerdrum posted:I watched a video gallery in tumblr bring a pair of E5-4620 V4's to 100% cpu utilization. the videos started playing again offscreen
|
|
#
?
Jun 14, 2017 07:33
|
|
|
Paul MaudDib posted:Matrix math tends to parallelize real well thanks to time-tested primitives like BLAS/LAPACK (which Numpy hooks as native C). I'm guessing if you're getting good scaling (4x speedup) on a 4-core already then you would get 8x out of Ryzen, yes. awesome thanks. and is there a sweet spot between the 1700, 1700x, and 1800x models?
|
|
#
?
Jun 14, 2017 07:37
|
|
|
Get the 1700 since there's very little difference between the chips other than the 1800X might clock 100MHz faster. If you live near a microcenter there are some good deals on 1700X/Mobo bundles as well.
|
|
#
?
Jun 14, 2017 07:44
|
|
|
shrike82 posted:awesome thanks. and is there a sweet spot between the 1700, 1700x, and 1800x models? The higher models have higher factory clocks and are binned better, so will OC slightly farther. But the difference is fairly minimal, all current Ryzen processors hit a wall at 3.9 GHz and even a good chip only make it to 4.0 or 4.1 GHz with a nearly dangerous amount of voltage. On the flip side the 1700 uses significantly less power, both stock and apparently also when OC'd for some reason. The X models have "XFR" which is like an auto-overclock thing (assuming you have sufficient cooling). But it only gives you another 100 MHz tops. The 1700 is cheaper than the other models and comes with a cooler included, which is possibly the biggest differentiator in the whole line (  ). IMO it's still the obvious choice in the Ryzen 7 lineup although less so than at launch, when the 1800X was like $200 more expensive than the 1700. If you want more factory speed then the 1700X or the 1800X are OK but really they don't offer anything you won't be able to live without.They all have the same L3 cache and stuff too, no difference there.
|
|
#
?
Jun 14, 2017 07:51
|
|
|
Mind you, if you're using your workstation for any sort of scientific computing, or computing where the possibility of an error is unacceptable to the finished product (note this isn't me saying that an error is unacceptable to the finished product, merely the possibility of one impugning your results), then you really don't want to overclock.
|
|
#
?
Jun 14, 2017 08:16
|
|
Cat Army 
|
Machine and deep learning are pretty resilient to imprecision (and inaccuracy) if you set up the models right. Anyway it's hobby stuff shrike82 fucked around with this message at 08:34 on Jun 14, 2017 |
|
#
?
Jun 14, 2017 08:31
|
|
|
shrike82 posted:Machine and deep learning are pretty resilient to imprecision (and inaccuracy) if you set up the models right. I was more thinking if you're running data for publication or something along those lines, but yeah go wild then.
|
|
#
?
Jun 14, 2017 09:04
|
|
|
Ryzen doing well in Dirt 4 source (german) (unless the game is GPU bound with a 1080ti) eames fucked around with this message at 13:22 on Jun 14, 2017 |
|
#
?
Jun 14, 2017 12:47
|
|
|
Posting from my new Ryzen 5 1600X + B350 build mothafuckas. This is pretty spiffy.
|
|
#
?
Jun 14, 2017 14:22
|
|
|
Wirth1000 posted:Posting from my new Ryzen 5 1600X + B350 build mothafuckas. This is pretty spiffy. Were you able to get your ram speeds squared away?
|
|
#
?
Jun 14, 2017 15:30
|
|
|
rex rabidorum vires posted:Were you able to get your ram speeds squared away? I'm at work right now and I went straight to bed last night when I finished building and installing Windows so I haven't had a chance to benchmark or speeds. The most the system has been on for is about 20 minutes last night and this morning but so far everything seems super stable. No errors on boot up or anything so quite happy. It's off right now as I wasn't gonna leave it on without being present. This is the RAM I ultimately went with: CMK16GX4M2A2400C16 Corsair Vengeance LPX 16GB DDR4-2400. It was in the QVL for the ASRock Fatal1ty AB350 Gaming K (which, by the way, has a layout that I absolutely detest) so I went with it. No issues whatsoever but once I get home from work I'm gonna do all sorts of stress tests and benchmarks to make sure everything is smooth and stable and running at advertised speeds.
|
|
#
?
Jun 14, 2017 15:42
|
|
|
I've had that exact ram in my Z170 system since Skylake was brand new. It's good stuff: does 2666 at stock voltage and timings so would probably run much faster with relaxed timings and a voltage bump, I've just never bothered to try.
|
|
#
?
Jun 14, 2017 15:46
|
|
|
I was worried about RAM, especially since nobody seems to use 32GB kits, but I just turned on XMP and it booted at 2666MHz. Also using a Corsair Vengeance LPX kit.
|
|
#
?
Jun 14, 2017 17:13
|
|
|
eames posted:Ryzen doing well in Dirt 4 I love it. One of the main reasons why frametime measurements were pushed forwards is because Crossfire setups had such terrible frame pacing. The measurement system designed to show how bad AMD was at delivering consistent frame rates is now showing off how AMD can deliver very consistent frame rates.
|
|
#
?
Jun 14, 2017 19:38
|
|
|
|
| # ? Apr 18, 2024 01:25 |
|
|
So the RAM is running at 2133, blargh.
|
|
#
?
Jun 14, 2017 21:57
|
|