|
12C/24T, 4.6GHz, and 105W TDP. Not bad. I wonder if that and R5 and lower tiers are where bum dies are going. If it's 3900X, not much room to go up in the 3,000 series. The Threadripper dream lives on!
|
# ? May 27, 2019 04:24 |
|
|
# ? May 17, 2024 07:48 |
|
Cao Ni Ma posted:Its supposed to be $400 unless they purposely mislead in the prestream Yeah I don't think they misled, I think they're doing Zen+ prices. Just completely didn't figure on them doing a 2 hour stream. edit: yeah $329-399, so above Zen+.
|
# ? May 27, 2019 04:25 |
|
Anarchist Mae fucked around with this message at 04:39 on May 27, 2019 |
# ? May 27, 2019 04:26 |
|
NewFatMike posted:12C/24T, 4.6GHz, and 105W TDP. Not bad. I wonder if that and R5 and lower tiers are where bum dies are going. If it's 3900X, not much room to go up in the 3,000 series. I can dream of a 64C/128T TR
|
# ? May 27, 2019 04:26 |
|
Sorry if these are spammy. Anarchist Mae fucked around with this message at 04:39 on May 27, 2019 |
# ? May 27, 2019 04:27 |
|
ratbert90 posted:I can dream of a 64C/128T TR 32C/64T was my ideal unit, so fingers crossed. Measly Twerp posted:
You provide a valuable service!
|
# ? May 27, 2019 04:27 |
|
So $79 for an extra 100 MHz and a higher pre-configured TDP limit on their "mainstream gaming processor"?
|
# ? May 27, 2019 04:28 |
|
lol RIP intel. $500 for more performance than a $1200 i9.
|
# ? May 27, 2019 04:28 |
|
3700X 8C/16T 3.6/4.4GHz 65W $329 3800X 8C/16T 3.9/4.5GHz 105W $399 3900X 12C/24T 3.8/4.6GH 105W $499 3900X is branded as Ryzen 9. Release date: 7/7
|
# ? May 27, 2019 04:29 |
|
Paul MaudDib posted:So $79 for an extra 100 MHz and a higher pre-configured TDP limit on their "mainstream gaming processor"? Its stupid, its just there to make the jump to the 3900x more palatable
|
# ? May 27, 2019 04:29 |
|
I was hoping the 12 core part was not the top of the stack for pricing reasons, oh well.
|
# ? May 27, 2019 04:29 |
|
Anarchist Mae fucked around with this message at 04:39 on May 27, 2019 |
# ? May 27, 2019 04:29 |
|
Are the R5 chips coming out at the same time or later?
|
# ? May 27, 2019 04:30 |
|
Anarchist Mae fucked around with this message at 04:40 on May 27, 2019 |
# ? May 27, 2019 04:31 |
|
Measly Twerp posted:Sorry if these are spammy. Maybe go back and convert them to timg but otherwise they're helpful
|
# ? May 27, 2019 04:31 |
|
This motherfucker carried his printer across town for Lisa to print her keynote speech, but they had a teleprompter
|
# ? May 27, 2019 04:31 |
|
Cao Ni Ma posted:Its stupid, its just there to make the jump to the 3900x more palatable unless it's binned substantially better... like the first-gen Ryzen processors.
|
# ? May 27, 2019 04:33 |
|
.
sincx fucked around with this message at 05:50 on Mar 23, 2021 |
# ? May 27, 2019 04:33 |
|
Yeah I meant more price after Intel responds. Feel like $400 isn’t really a tough ask price cut wise, the 9900k has gone to like $460 already. A real pricewar might break out with equivalent parts and that is fantastic for us.
|
# ? May 27, 2019 04:34 |
|
OhFunny posted:3700X 8C/16T 3.6/4.4GHz 65W $329 I am just so confused about the such minor differences between the 3700x and 3800x. I also wonder if that is an all core boost number, but probably not since I would assume they would specify a larger number if it could reach it even if it was only a single core. Now to wait and find out the overclocking potential of each.
|
# ? May 27, 2019 04:36 |
|
Broose posted:I am just so confused about the such minor differences between the 3700x and 3800x. I also wonder if that is an all core boost number, but probably not since I would assume they would specify a larger number if it could reach it even if it was only a single core. Now to wait and find out the overclocking potential of each. That has to be binning.
|
# ? May 27, 2019 04:37 |
|
AMD wanted to go head to head with the 9900k in gaming, and looks like they needed to crank the TDP to get over the hump. I am disappointed that a 16core doesn’t look to be in the cards based on the naming scheme. Even with a crazy price or TDP, woulda been a fun part. I imagine AMD wants to keep a market for better margin TR4 parts, and Intel only has 10 cores on the horizon for desktop, but it’s still disappointing.
|
# ? May 27, 2019 04:43 |
|
ratbert90 posted:lol RIP intel. $500 for more performance than a $1200 i9. I mean, for multithreaded applications you could already get that in a 1950X. It's not like Intel had a strong position in the workstation segment. Then 3900X is definitely top of the line in its segment but like most 12-cores a little awkwardly positioned. It's not much better than a 6 or 8 core for non-workstation usage, yet still has too few cores to be a very impressive workstation. I was rather hoping for a high clockspeed 16-core as a superior alternative to the mid-range Threadrippers; presumably AMD can and possibly will do that but are not currently that keen on cannibalizing their mid-range Threadrippers.
|
# ? May 27, 2019 04:48 |
|
The stack is setup like that so when they introduce the 16c they can drop the 12c price and get rid of one of the 8 core skus. Like gen 1 ryzen, where all the new 1800x parts are just branded 1700x now. Also setup to get free money from people who want the best performing CPUs. 12c price is relatively fair vs $330 8c. You pay a $5 premium for double the cache and the same per-core price outside of that. Khorne fucked around with this message at 04:57 on May 27, 2019 |
# ? May 27, 2019 04:55 |
|
Quick question regarding cache: The 9900K has 16MB of cache while the Ryzen 9 3900X has gently caress you 70MB... does cache make a significant difference for gaming?
|
# ? May 27, 2019 04:58 |
|
Otakufag posted:Quick question regarding cache: The 9900K has 16MB of cache while the Ryzen 9 3900X has gently caress you 70MB... does cache make a significant difference for gaming? Remains to be seen how its integrated and applied (if there are issues accessing the cache). Historically a huge cache like on the 5th core series has been great for frametime (1% and 0.1%) performance to make gameplay smoother.
|
# ? May 27, 2019 05:02 |
|
Cygni posted:I am disappointed that a 16core doesn’t look to be in the cards based on the naming scheme. Even with a crazy price or TDP, woulda been a fun part. I imagine AMD wants to keep a market for better margin TR4 parts, and Intel only has 10 cores on the horizon for desktop, but it’s still disappointing. There's no reason we can't get a 3990wx down the line.
|
# ? May 27, 2019 05:02 |
|
Ragingsheep posted:Are the R5 chips coming out at the same time or later? I believe those are being unveiled at E3. Khorne posted:The stack is setup like that so when they introduce the 16c they can drop the 12c price and get rid of one of the 8 core skus. Like gen 1 ryzen, where all the new 1800x parts are just branded 1700x now. I agree with this. Dr. Su did announce the 3900X as part of the new R9 "family". One CPU isn't a family. I wonder what the naming would be. 3950X?
|
# ? May 27, 2019 05:02 |
|
Cygni posted:I am disappointed that a 16core doesn’t look to be in the cards based on the naming scheme. Or maybe it turns out it didn't work or there's not enough 8s at all.
|
# ? May 27, 2019 05:03 |
|
Otakufag posted:Quick question regarding cache: The 9900K has 16MB of cache while the Ryzen 9 3900X has gently caress you 70MB... does cache make a significant difference for gaming? They even out. There should certainly be some tasks where the better latency of the 9900k wins and achieves higher ipc, and there should be other tasks where the massive cache wins. It depends what's going on. In most cases, they should be fairly balanced for gaming. For productivity, most things benefit more from the cache. edit: I removed some speculation because the ipc difference varies so much. AMD should lead by ~3% ipc in most games if their 15% number is correct. In certain tasks they will lead by as much as 19% if it's correct. In certain latency-bound games or tasks Intel should lead by as much as 8%-12% ipc. Intel has better clock speeds so it's likely the 9900k still out performs zen2 in most games if 4.6 is the ceiling. The massive cache should help a lot in any game with intense pathfinding or highly optimized tasks that can reside largely in cache. I can't think of any specific titles here that people are trying to push frames in, but it should slay the benchmark rts games. OhFunny posted:I agree with this. Dr. Su did announce the 3900X as part of the new R9 "family". One CPU isn't a family. I wonder what the naming would be. 3950X? Khorne fucked around with this message at 05:32 on May 27, 2019 |
# ? May 27, 2019 05:04 |
|
Two Ryzen 5's are in AMD's press release, AnandTech has some details on them: https://www.anandtech.com/show/14407/amd-ryzen-3000-announced-five-cpus-12-cores-for-499-up-to-46-ghz-pcie-40-coming-77 Friends don't let friends buy the 3600X.
|
# ? May 27, 2019 05:08 |
|
I'm eager to see benchmarks of the 3600.
|
# ? May 27, 2019 05:16 |
|
My WAG is the huge TDP boost for the 3700X and the 3600X over their non-X counterparts is to allow for XFR to do its thing. If they end up auto-OC'-ing all core to around their boost frequencies and stay near the actual listed TDP with the stock HSF than that actually isn't a bad deal. No 16C, prices are a little higher than I'd want, and the clocks are little disappointing to me but otherwise its really pretty far from being bad at all.
|
# ? May 27, 2019 05:18 |
|
I'd really love a 3900 (No X) running at 3.2-3.4GHz at 95W. But if the 3900X is all there is, then that's what I'll be migrating my compute stack to. I don't know if 70MB of L3 will mean a lot for gaming, but it'll sure mean a lot for scientific computing. One of the projects I crunch for uses the Rosetta suite, and it wants to keep 4MB of data resident in L3 at all times, per running thread. If you're low on cache and consistently have misses, that thread's runtime goes up by a factor of about 2.5X (~4500s to ~11500s on my hardware). So yeah. It's huge for non-gaming applications, for sure. This puts the 3900X on par with high-end Xeons. mdxi fucked around with this message at 05:30 on May 27, 2019 |
# ? May 27, 2019 05:28 |
|
Otakufag posted:Quick question regarding cache: The 9900K has 16MB of cache while the Ryzen 9 3900X has gently caress you 70MB... does cache make a significant difference for gaming? Yes. It’s very important to optimize games to reduce the frequency of what is called a "cache miss". That is trying to load data from the cache that isn’t there, which requires the CPU to fetch it from main ram which introduces huge latency, and thus frame rate loss. xgalaxy fucked around with this message at 05:34 on May 27, 2019 |
# ? May 27, 2019 05:29 |
|
mdxi posted:I'd really love a 3900 (No X) running at 3.2-3.4GHz at 95W. But if the 3900X is all there is, then that's what I'll be migrating my compute stack to.
|
# ? May 27, 2019 05:32 |
|
I was hoping for a little higher clocks but the IPC gain is more than I expected so it evens out, will definitely be picking up the 3900X at some point.
|
# ? May 27, 2019 05:35 |
|
So basically only the 3900X, 3700X and 3600 matter so far, with the 3800X being yet another trap like the 1800X. No 3500 likely and will get the 3400G, 3200G (maybe a 3100G, 6CU Vega?) and obviously the Athlons. Maybe an R9 3950X 16C/32T in November? Also surely these things have to have some headroom over the core boost? Otherwise they're going to run really cold if they tap out around 105-125W. GN was talking about them pulling up to 300W on air, so maybe they can legit hit 4.8Ghz or so but the voltage curve is nuts. I don't think it's reaching to say AMD doesn't want to sell higher clocking CPUs that go crazy on the TDP. Possibly Zen2+ or Zen3 on 7nmEUV will have a better voltage curve? Also if the RX 5700 matches the RTX 2070 sounds like the RX 5800 matches the RVII.
|
# ? May 27, 2019 05:37 |
|
Khorne posted:You can underclock to hit the power curve sweet spot. An excellent suggestion. Fun times ahead with BIOS settings and the killawatt.
|
# ? May 27, 2019 05:38 |
|
|
# ? May 17, 2024 07:48 |
|
EmpyreanFlux posted:I don't think it's reaching to say AMD doesn't want to sell higher clocking CPUs that go crazy on the TDP. EmpyreanFlux posted:Possibly Zen2+ or Zen3 on 7nmEUV will have a better voltage curve? TSMC's 7FF+ looks to be a bit better than their current 7nm but I wouldn't expect miracles from the process change alone. Maybe a couple hundred more Mhz along with a nice price cut given the smaller die size too? AMD would have to work some significant architectural improvements to really make a difference to clockspeeds I'd think and the general trend seems to be to focus more on improving IPC over clockspeeds so I kinda doubt they will.
|
# ? May 27, 2019 06:05 |