Delusibeta posted:So, the impression I'm getting is that Ryzen 7 effectively renders Intel's X99 platform moot, but lags noticeably behind Intel's mainstream chips when it comes to gaming, and that the Core i5 7600k remains the default CPU choice for gaming builds (for now). That seems accurate, though it's hard to say what gaming performance will end up being once all the kinks with motherboard UEFI and so on are worked out. I think AMD should have launched a week or so later, maybe the mobo makers would have had their stuff in line for launch that way.
|
|
# ? Mar 3, 2017 18:25 |
|
|
# ? May 21, 2024 14:28 |
|
Dante80 posted:More or less. Remember, that it is also Intel that renders the X99 moot (for new builds of course), since they are changing sockets anyway soon. Ultimately, it does come down to pricing, but considering the bottom end Ryzen 7 has the same US price as the i7 7700k, the only way is down. I am looking forward to whatever price Ryzen 3s wind up being, since getting 80% of the performance of the 7600k for half the price seems like a very fair trade to me.
|
# ? Mar 3, 2017 18:26 |
|
Delusibeta posted:Ultimately, it does come down to pricing, but considering the bottom end Ryzen 7 has the same US price as the i7 7700k, the only way is down. I am looking forward to whatever price Ryzen 3s wind up being, since getting 80% of the performance of the 7600k for half the price seems like a very fair trade to me. The old leaked prices were 260$ for the top 6c/12t, 200$ for the top 4c/8t and 150$ for the top 4c/4t. The last two get half the L3 cache, since one of the two CCXs is disabled. On the other hand, this potentially means more overclocking and no multithread fuckery (the problems that we see now with the two CCXs working together in R7). Sadly, the R5s are Q2 and the R3s H2 (probably Q3). Dante80 fucked around with this message at 18:38 on Mar 3, 2017 |
# ? Mar 3, 2017 18:35 |
|
Delusibeta posted:So, the impression I'm getting is that Ryzen 7 effectively renders Intel's X99 platform moot, but lags noticeably behind Intel's mainstream chips when it comes to gaming, and that the Core i5 7600k remains the default CPU choice for gaming builds (for now). It only renders X99 moot if you don't need lots of memory bandwidth (starting to look like an issue for Ryzen 8C16T) or SIMD throughput (advantage Intel) or lots of PCIe lanes (probably not an issue for most). Basically, you need to examine your workload carefully and see if the Ryzen platform's shortcomings matter. As others have noted, X99 is probably going to be replaced anyway so it's hard to say right now.
|
# ? Mar 3, 2017 19:02 |
|
TastyPC is back. https://www.youtube.com/watch?v=2brYnTGTSR8
|
# ? Mar 3, 2017 19:19 |
|
Also, a very interesting meta video about the reviews, also has some official AMD correspondence inside. https://www.youtube.com/watch?v=TBf0lwikXyU
|
# ? Mar 3, 2017 19:30 |
|
Munkeymon posted:A 7700K wouldn't be enough? That was going to go into my new desktop this year but I'm at least considering AMD now because it'd be nice to have a bit of a guarantee that I'll be able to swap out the CPU when they get better in a couple years. 7700k is still a lovely rear end 4core cpu that'll hang itself hard if I let 4 threads of x264 at 60fps at it while playing games though?
|
# ? Mar 3, 2017 19:36 |
|
fishmech posted:We have already seen competition with Intel multiple times. It never made Intel drop their needlessly expensive super high price chips for being on sale for $1000 or more. Remember we are talking about the chips that are only marginally better than the normal top of the line, but take hundreds of dollars higher prices. They've consistently been on sale since the 90s. Exactly. It's such an absurd, over-the-top purchase that the category is price inelastic. I was at a Nissan dealership the other day (looking at a used minivan ) and I was chatting with one of the sales dudes about the GT-R they had in the dealership. This is a car with an MRSP starting at $110k or $180k depending on the package. I asked him when was the last time they sold one, and he told me that their most recent sale of that model was a few months ago. A guy flew in to AZ from Oregon to buy it, because it was the configuration in the right trim package/color/whatever that he wanted, and he paid a fortune to ship it back to his home in Oregon. That guy was a GT-R buyer, buying a needlessly expensive halo product sports car in a world where there are much better values in sports cars.
|
# ? Mar 3, 2017 19:41 |
|
Truga posted:7700k is still a lovely rear end 4core cpu that'll hang itself hard if I let 4 threads of x264 at 60fps at it while playing games though? I wouldn't think one video encoding thread would occupy a whole core (it's 4c/8t), but I also haven't looked into it.
|
# ? Mar 3, 2017 19:50 |
|
Boiled Water posted:The two black plastic bits, if they were removed and the holes put closer to the socket you should be able to just squeeze something in there. I bet that volume is reserved for the heat sink in the spec
|
# ? Mar 3, 2017 19:55 |
|
Truga posted:7700k is still a lovely rear end 4core cpu that'll hang itself hard if I let 4 threads of x264 at 60fps at it while playing games though?
|
# ? Mar 3, 2017 20:01 |
|
I can't play a game and watch a video at the same time with my FX-8350 without the video stuttering. Even DVDs skip, lol
|
# ? Mar 3, 2017 20:09 |
|
Dante80 posted:TastyPC is back. Uploader comment says they don't have a retail chip, it's an engineering sample.
|
# ? Mar 3, 2017 20:11 |
|
Nice launch. golem.de posted:The MSI board was delivered with BIOS version 113, until last Friday a new one appeared.
|
# ? Mar 3, 2017 20:13 |
|
Truga posted:7700k is still a lovely rear end 4core cpu that'll hang itself hard if I let 4 threads of x264 at 60fps at it while playing games though? Source your quotes.
|
# ? Mar 3, 2017 20:30 |
|
Munkeymon posted:I bet that volume is reserved for the heat sink in the spec Very likely, but it looks like a decent amount of space is wasted compared to the good ol' LGA mounts.
|
# ? Mar 3, 2017 20:32 |
|
Munkeymon posted:I wouldn't think one video encoding thread would occupy a whole core (it's 4c/8t), but I also haven't looked into it. Video encoding seems like it would be exactly the kind of workload that would saturate a core's execution units. Not a lot of dependencies and branches, just lots and lots of math. E: I wonder if it would be possible to have a flag that declares that one thread is higher priority than the other so that it's instructions get preferential treatment. I'm thinking in most games you're doing a lot of pointer chasing and waiting on cache misses so you might have time to let an encoding thread work, but when the game thread can run you'd want it to start going ASAP. Does that seem reasonable or am I crazy Yaoi Gagarin fucked around with this message at 21:10 on Mar 3, 2017 |
# ? Mar 3, 2017 21:04 |
|
Not gonna lie, part of why I want Ryzen is to use the -threads 16 flag with ffmpeg.
|
# ? Mar 3, 2017 21:07 |
|
https://www.youtube.com/watch?v=nsDjx-tW_WQ Dude with the obvious GPU bottlenecking redid testing on 720p low instead of 1080 ultra.
|
# ? Mar 3, 2017 21:08 |
|
Wait do people actually pronounce it as "Bi-ahhs"? I've only ever heard it pronounced as "Bi-o's"
|
# ? Mar 3, 2017 21:16 |
|
I was just thinking about that as well. I see it as "Bio" with an s on the end, when I'm pronouncing it.
|
# ? Mar 3, 2017 21:18 |
|
buglord posted:Wait do people actually pronounce it as "Bi-ahhs"? I've only ever heard it pronounced as "Bi-o's" Probably the same people that pronounce Ohio as like ohiah.
|
# ? Mar 3, 2017 21:20 |
|
Alereon posted:So to clarify, you're trying to use software to compress the video for the game you're playing in real-time? Yeah that'll be hard on a quad-core CPU but it seems like there's better solutions, like hardware encoding. GPU encoding produces poo poo quality unless you're willing to stream at higher bitrates and then everything goes to poo poo anyway, because youtube will recompress your stream to poo poo so people can actually watch it. GPU encoding is great for dumping videos to disk, since there's basically no performance hit, and you can just give it a 30Mb/s bitrate. Internet streaming not so much, and at 5Mbit/s, which I stream at (max for twitch is 3.5, even, which is why I moved to youtube), x264 produces a vastly better stream that youtubes digest far better. I just have to stream at 30fps currently because streaming at 60 and trying to retain the quality occupies the 4 cores I have and then the game runs at <30 fps and there's no point. Ryzen is basically the cpu that was made specifically for my use case, as far as gaming goes. I was considering buying a second pc to encode on exclusively (still 50% cheaper than one of the decent stream encoding cards), but I now I can sell my old poo poo and get a brand new pc with that cash that'll stream smoothly and also have a new gpu I can VR with. That said, I'm still not buying anything until these bugfixes get benchmarked and also VR benchmarks. If ryzen is somehow super poo poo for vr, eames posted:Nice launch. Yeah, this is pretty funny Truga fucked around with this message at 21:42 on Mar 3, 2017 |
# ? Mar 3, 2017 21:39 |
|
servethehome have posted their 1700 benchmarks. They've also managed some non-trivial bumps in their updated 1700x numbers by using a newer Linux kernel.
Pablo Bluth fucked around with this message at 21:58 on Mar 3, 2017 |
# ? Mar 3, 2017 21:45 |
|
Ooh, a new site I've never heard of that sounds semi-interesting. Bookmarked for later!
|
# ? Mar 3, 2017 21:52 |
|
BurritoJustice posted:When Ryzen loses it really loses. It gets beaten by 40FPS by a stock 7700K in The Witcher 3. This result in particular has been loving with me. It's just too bad to believe. I mean, on par with an FX-8320? This video makes me think there was something wrong with that site's testing methodology or equipment. The results overall seem more consistent with what you'd expect, given the IPC/clock speed disadvantage, i.e. Ryzen is on average about 10-15% slower. In fact, while Ryzen's gaming performance clearly isn't as fast as a 7700K, that video makes me question whether it's really going to be a serious tradeoff at all. If it's a small unnoticeable hit in gaming performance in exchange for brutally faster encoding/compiling performance, to me that's an easy decision. NIGARS fucked around with this message at 22:20 on Mar 3, 2017 |
# ? Mar 3, 2017 22:17 |
|
Truga posted:GPU encoding produces poo poo quality unless you're willing to stream at higher bitrates and then everything goes to poo poo anyway, because youtube will recompress your stream to poo poo so people can actually watch it. [H] did some VR stuff in their review, it looks stronger in VR than in the general gaming benchmarks. http://www.hardocp.com/article/2017/03/02/amd_ryzen_1700x_cpu_review
|
# ? Mar 3, 2017 22:24 |
|
Pablo Bluth posted:servethehome have posted their 1700 benchmarks. They've also managed some non-trivial bumps in their updated 1700x numbers by using a newer Linux kernel. This is looking really good. AMD should have a solid server product when Naples lands, I'm really optimistic.
|
# ? Mar 3, 2017 22:27 |
|
NIGARS posted:This result in particular has been loving with me. It's just too bad to believe. I mean, on par with an FX-8320? It's the latter, a couple games have unexplained swings. You can see the tradeoffs below, be sure to interact with the charts to select/deselect your games/apps. https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/3/ https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/4/
|
# ? Mar 3, 2017 22:35 |
|
NIGARS posted:This result in particular has been loving with me. It's just too bad to believe. I mean, on par with an FX-8320? It is impossible for Ryzen to be as fast in games as the 7700k. There is a sizable clockspeed delta (OCed or stock for both platforms) and also a smaller IPC difference. The only way for Ryzen to do better would be in games that can leverage more than 8 threads properly. But here comes the disparity. In most benches concerning gaming, Ryzen is doing even worse than what we expected pre-launch. In other words, the benchmarks do not reflect the raw computing power the CPU shows elsewhere, even in a per core basis (when you do compare it in gaming benchmarks to haswell/broadwell-E). And it fails to surpass the 7700k even in games that are marketed for utilizing more than 8 threads properly. There are a number of reasons for that, and much more speculation. And much of that can be attributed to things pertaining to the terrible (in PR terms) launch, that came out a week or two too soon. Some though cannot, and AMD might need a new revision to change them. We will see.
|
# ? Mar 3, 2017 22:44 |
|
Twerk from Home posted:This is looking really good. AMD should have a solid server product when Naples lands, I'm really optimistic. Reading this, the Uarch is poised to deliver crushing efficiency in the 30-45w department. And the MCM approach to Naples (with its octuple channel memory and the shitload of lanes it provides in 32c ss and 48c/64c ds) might be a perf/$ winner. Whether Opteron is back or not though, remains to be seen. Intel is also dropping a new server part.
|
# ? Mar 3, 2017 22:48 |
|
Dante80 posted:Reading this, the Uarch is poised to deliver crushing efficiency in the 30-45w department. And the MCM approach to Naples (with its octuple channel memory and the shitload of lanes it provides in 32c ss and 48c/64c ds) might be a perf/$ winner. This will hopefully also translate into some really good laptop chips.
|
# ? Mar 3, 2017 22:53 |
|
Twerk from Home posted:This is looking really good. AMD should have a solid server product when Naples lands, I'm really optimistic. wrong post MaxxBot posted:[H] did some VR stuff in their review, it looks stronger in VR than in the general gaming benchmarks. Maybe, although the fact that it lags behind the 2600K in about half the tests (2600K had 500 MHz or 12% clock advantage) is slightly concerning. For now I'm chalking it up to BIOS/EFI since apparently the ASUS board had the most problems among the review samples. kirtar fucked around with this message at 23:01 on Mar 3, 2017 |
# ? Mar 3, 2017 22:55 |
|
Arzachel posted:This will hopefully also translate into some really good laptop chips. Yes, it might. AMD is pushing a lot of stuff this year in all markets, they might have a winner somewhere. kirtar posted:Maybe, although the fact that it lags behind the 2600K (2600K had 500 MHz or 12% clock advantage) is slightly concerning. For now I'm chalking it up to BIOS/EFI since apparently the ASUS board had the most problems among the review samples. Lags in what task relevant to the server market? Productivity per $ and per watt is what governs it. Dante80 fucked around with this message at 22:58 on Mar 3, 2017 |
# ? Mar 3, 2017 22:55 |
|
If my main use case is 11-boxing eve online, actively running another game, and watching some sort of streaming video (twitch, amazon, youtube), would ryzen be a better purchase over an intel hex-core with the information out so far? For reference, I am currently running a 4690k more or less at stock, and I have to stop doing at least one of the things above to approach reasonable performance. Each eve client takes between 4 and 15% cpu in task manager depending on if they're backgrounded (limited to 10fps), my second game usually takes 50%, and streaming video can take a lot especially from amazon. This squarely basically pegs me at 100% load at all times. I'm not in the biggest rush to buy, but I've been wanting to upgrade for ages because the 4690k felt like a downgrade from the 980x I had before it so skylake-x might be a little too far out to wait. I could overclock a bit, but I doubt that will get me where I want to be
|
# ? Mar 3, 2017 22:57 |
|
Dante80 posted:Yes, it might. AMD is pushing a lot of stuff this year in all markets, they might have a winner somewhere. I seem to have quoted the wrong post. I was meaning to quote the VR gaming thing posted by MaxxBot MaxxBot posted:[H] did some VR stuff in their review, it looks stronger in VR than in the general gaming benchmarks.
|
# ? Mar 3, 2017 22:58 |
|
kirtar posted:I seem to have quoted the wrong post. I was meaning to quote the VR gaming thing posted by MaxxBot No, its me. I suck at everything.
|
# ? Mar 3, 2017 22:59 |
|
Dante80 posted:No, its me. I suck at everything. Nah, I definitely quoted the wrong post. I edited the right one in just in case someone else runs into it reading top to bottom.
|
# ? Mar 3, 2017 23:00 |
|
Cool overclocking video. https://www.youtube.com/watch?v=IHjCVsJdvZw
|
# ? Mar 3, 2017 23:16 |
|
|
# ? May 21, 2024 14:28 |
gwrtheyrn posted:If my main use case is 11-boxing eve online, actively running another game, and watching some sort of streaming video (twitch, amazon, youtube), would ryzen be a better purchase over an intel hex-core with the information out so far? Well.... Yeah, of course it would feel like a downgrade. You went from a 12-thread cpu to a 4-thread one. In the simplest terms, you basically went from being able to run everything over 2-3 clock cycles (again, highly simplified) to having to wait 4-5 clock cycles for the same thing on the i5.
|
|
# ? Mar 3, 2017 23:23 |