|
Dr. Fishopolis posted:There are plenty of amazing thin and lights, I dunno what you're waiting for. Ice Lake is gonna be a decent speed bump but the IGP is still worse than an MX150. If you're waiting for an ultrabook that can do current games at 1080p, you're gonna be waiting forever. Get your bro a laptop, be a bro. Won't be waiting that long at all; the 3.2 pound blade stealth with a 1650 should be out in a few months.
|
#
¿
Sep 21, 2019 16:42
|
|
|
|
| # ¿ May 22, 2024 08:56 |
|
|
Dr. Fishopolis posted:Well let's hold out for the thermals on that one, but I'm cautiously optimistic. I'm the market segment; I'm willing to do medium settings, but the blade 14 I have now is big enough that it annoys me. I want something that I can game on, use comfortably in bed, bring with me wherever and whenever without thinking about it and I'm willing to pay what that combination of portability and power costs. It's a niche market for sure, but it does exist. Edit: since this is the amd thread, I was also hoping for a hypothetical 14cu navi / zen2 based apu with hbm that could fill the ultralight gaming niche as well, but it doesn't seem likely that that's going to exist 
VorpalFish fucked around with this message at 18:07 on Sep 21, 2019 |
|
#
¿
Sep 21, 2019 17:57
|
|
|
Budzilla posted:I noticed that too. I'm just annoyed that Intel finally says 'look a great iGPU' in TYOOL 2019 and throws out samples in configurations that won't be sold. Complaints about poo poo performance from Intel graphics has been around for years and no all of a sudden they care. I don't think it's correct to say this configuration won't be sold; every announced ice lake laptop I've seen claims to be using it (acer swift, dell xps, blade stealth). I don't doubt there will be skus using awful ram, but it looks like you will be able to buy dual channel lpddr4 3733 in actual retail laptops.
|
|
#
¿
Sep 22, 2019 15:17
|
|
|
AIOs make sense when: 1) you have height restrictions due to your case preventing you from using the really effective tower coolers 2) you move your computer a lot (often goes with 1) 3) GPUs where your aftermarket air options are basically the accelero and morpheus 2 and make your card 3 slots 4) cases with heavily restricted airflow in the main chamber but better potential airflow at radiator mounting points (something like the evolv shift) Otherwise big air coolers are both better at cooling value and at noise normalized performance and should be the default choice.
|
|
#
¿
Dec 15, 2019 00:07
|
|
|
Leandros posted:I have an i5 750 so it's time for a refresh and AMD definitely seems like a smart bet this time. I'd prefer to do it sooner rather than later, but the ~20% IPC gains the Zen3 apparently touts is sounding pretty good too. Would it make sense to go for a 3700X now with an X570 mobo and swap out the CPU for a 4k series somewhere down the line or will the X570 be a bottleneck? 3800X is about 35 euros more but doesn't seem worth the investment. I'm not getting any super authoritative info that Zen3 would even run on AM4, aside from the fact that Epyc3 will be running DDR4 and a move to AM5 would imply DDR5 memory, which won't be manufactured in high enough volumes around the release of Zen3. Considering the X570 is currently high-end, I would assume it ought to be enough, but I honestly don't know enough about this stuff. There's always something better around the corner with computer hardware. You should upgrade when you want something better unless you know the newer architecture is releasing like next month. I'd also forget about upgrading zen2 to zen3. Socket compatibility sounds great in theory but in practice, spending 300+ for a 15-20% performance uplift is terrible value. Coming from a 750, you'd probably be super happy with a 3700x or even a 3600 depending on your workloads. Also consider b450 over x570 if you don't have a compelling use case for pcie4 because it's cheaper and gently caress active chipset cooling forever.
|
|
#
¿
Dec 29, 2019 18:05
|
|
|
Leandros posted:I'd obviously sell the 3700X if I were to get a new one, so it'd not be that big a hit, but you're probably right that it might not be worth it. As for X570, I was originally looking for a mobo with 8 SATA ports as I have a buttload of storage and those seemed the cheapest option. I'm now going for a NAS build, so the more common 6 SATA ports should be fine. I mean a 2080ti right now is fine on 8 lanes of pcie3, let alone 16. I don't see pcie4 being useful for gpus for awhile. Better use case is like 10/100gig ethernet or feeding nvme drives with fewer lanes so you can provide more connectivity from the chipset, or maybe some future thunderbolt spec. *yes I know those recently released amd 5500s are showing bottlenecks on pcie3 x8 probably because amd hosed something up but they're garbage value cards and nobody should be buying them anyways.
|
|
#
¿
Dec 29, 2019 20:54
|
|
|
movax posted:I think it’s more appealing for the first reasons there, using fewer lanes / physical I/O where possible. I always liked doing bandwidth bridging in designs with PCIe Gen 2 or 3 feeding into a switch that would fan out to slower devices that didn’t need all the bandwidth. Getting 10 GbE in x1 would be great for density. 100% agreed, but in think we're a long ways away from the point where a typical home user is going to benefit which is why I tend to steer people away unless they have a specific use case. How many people need even a single 10Ge in their home desktop at this point, or will even in the next 5 years?
|
|
#
¿
Dec 29, 2019 21:31
|
|
|
I mean you're usually going to get way better returns buying the fastest single gpu you can afford for a pure gaming workload so one would assume if you're considering a 9900k/s over the 3700x for gaming you're pairing it with a 2080s at least or you've got like an emulator use case that really favors intel or something. Doesn't make a ton of sense to spend +$150 on a processor when that could buy you a faster GPU.
|
|
#
¿
Apr 3, 2020 20:31
|
|
|
Paul MaudDib posted:the price to performance curve falls off a cliff after the 2070S, a 2080S is spending 60% more for 13% more performance. If you're willing to spend an extra $300 for a 15% return on your graphics, why not your CPU, which will last you multiple GPUs? Unless you live in a region with very different pricing from me, its like a 40% jump in cost, not 60%. I'm not saying it's not poor value, stuff at the far end of the performance curve usually is. But 3700x -> 9900k is also bad value. If I'm willing to spend into diminishing returns for better absolute performance and my use case is gaming (and I can only afford one), I'm taking the 2080s every time. Obviously there's a point where it's ridiculous which is why I didn't say ti - at that price point the difference in processor cost doesn't even get you halfway to the next gpu step Edit: I guess it's only fair to mention the 9700k which gets you the same performance in games at least today as the 9900k at 85% of the cost. I could see pairing that with less than a 2080s if you're willing to risk "only" having 8c8t. VorpalFish fucked around with this message at 22:12 on Apr 3, 2020 |
|
#
¿
Apr 3, 2020 21:51
|
|
|
If you really want low maintenance you can go with one of the graphite sheet products as your TIM instead of paste and that basically shouldn't degrade at the cost of slightly worse up front performance.
|
|
#
¿
Apr 14, 2020 16:11
|
|
|
The 4 core zen2 parts are actually probably faster for most games so I can see buying one for that purpose as a placeholder with the intent to upgrade to a 6/8c zen3 part when it launches. 1600af is obviously the better choice for long term use, but afaik at least now no games are choking on 4c8t, it's 4c4t that's become an issue.
|
|
#
¿
Apr 22, 2020 13:16
|
|
|
I mean compared to what throwing them out it's all profit. But the cost of production and r&d hasn't really changed so it's not like these cost less to make than a 3900x or whatever.
|
|
#
¿
Apr 24, 2020 15:06
|
|
|
To be fair, they probably *will* be better at games at least for the immediate future, which is why I thought one of the potential use cases is someone building a gaming pc now but who knows they'll want to upgrade to zen3 which AMD says is coming later this year. 3100 now, then drop in a hypothetical 4600/4700(x) in November doesn't sound so bad. Edit: although I guess if you were willing to pay 9900k money for probably 9900k gaming performance you could just... buy a 9900k now so I guess it's a dumb idea after all. VorpalFish fucked around with this message at 18:50 on Apr 24, 2020 |
|
#
¿
Apr 24, 2020 18:44
|
|
|
Evil Robot posted:This is probably a dumb question; but is it the case that generally Ryzen 3000 CPUs run hot? At stock on a Ryzen 3900X on a Asrock B450M Pro4 with a NH-U12S cooler, F@H peaks at around 90C. With the stock cooler it would peak at 95C. That sounds way high. Do you have pbo turned on / no airflow / 40c ambient temps.
|
|
#
¿
Apr 25, 2020 19:57
|
|
|
I guess you could try remounting and repasting the heatsink, but 50 over ambient in prime 95 with that CPU and cooler doesn't sound like you're making bad contact. Are holding pretty steady at the 140w power limit in prime?
|
|
#
¿
Apr 26, 2020 01:44
|
|
|
If I'm reading some of the reviews right it looks like the 3300x might be getting better 1% lows than the 3100 even with the 3100 overclocked. Looks like there might be something to having all the cores on one ccx.
|
|
#
¿
May 7, 2020 20:09
|
|
|
gradenko_2000 posted:I have a bit of a historical/technical question: given that the Bulldozer architecture's "modules" were sharing a single floating-point unit, which is an argument for why they weren't really "cores" per se, and supposedly contributes to their lackluster performance compared to its Intel contemporaries, what made AMD think that that was a good idea? I believe their thinking was that integer workloads were much more common in consumer space, so a module would behave like 2 cores most of the time. And while it is an example of stretching the truth, I don't think that shared logic was the one thing holding the construction cores back; they were slow at everything including single threaded performance where the shared fpus shouldn't matter at all. I think it would be wrong to frame the failings of the uarch as a single bad bet by amds design team. Edit: it also probably would have worked better with earlier os support; knowing for example to schedule on cores 0,3,5,7 then 1,2,4,6 or whatever. VorpalFish fucked around with this message at 14:49 on May 8, 2020 |
|
#
¿
May 8, 2020 14:44
|
|
|
EmpyreanFlux posted:The 10900K reviews are like "This roughly competes with a 3900X!" but like, 3900X not only is cheaper but it's not a power hog running at 90°C to do so. It's not a bad processor per se but I feel like every review seems to have an element of "This is fine" dog in it. The 10700k and the 10600k are reasonable choices if your use case is high refresh rate gaming, you pair with a very fast gpu, and you care more about absolute performance than perf/dollar and perf/watt. Otherwise, probably not so much.
|
|
#
¿
May 21, 2020 18:51
|
|
|
I mean yes, they're terrible value in the current market so just, you know don't buy them. Like the 3600/3700x/3900x still exist and are great value for their respective use cases.
|
|
#
¿
Jul 9, 2020 19:55
|
|
|
PBO and overclocking on zen2 is mostly pointless. Spend your time tuning memory frequency and timings instead you'll get way more out of it.
|
|
#
¿
Jul 13, 2020 14:06
|
|
|
I would pay more for b550 than a comparably specced x570 just to avoid even the possibility of the fan spinning so I'm glad there are options.
|
|
#
¿
Jul 13, 2020 20:11
|
|
|
NewFatMike posted:If you're looking for a no compromises build, you're not using mITX, hth If you don't use itx, you're compromising on volume. 
|
|
#
¿
Jul 16, 2020 17:52
|
|
|
Paul MaudDib posted:I think they're going to go even lower and that may pose a problem for the new chips. I doubt zen2 will coexist with zen3 as a lower cost alternative the way zen+ has with zen2 since they're likely fabbed on the same process and there's not exactly a lot of capacity available on tsmc7. Doesn't seem worth it to port to another process either.
|
|
#
¿
Jul 29, 2020 17:59
|
|
|
Afaik they've only committed to launching sometime in 2020. B550 and x570 boards will support zen3 though so if you're in a hurry you could grab a 3600 and drop in an upgrade in a few months as long as you don't mind lighting like $160 on fire (probably can recoup some of that on eBay or whatever).
|
|
#
¿
Aug 27, 2020 21:27
|
|
|
Mr.Radar posted:When AMD shipped X570 they had to hastily repurpose their Epyc I/O die as a chipset because ASMedia (their normal chipset partner) didn't have PCIe 4 ready yet. ASMedia has probably caught up now (B550 was ASMedia again) so an X590 chipset could just be a lower-power remake of X570 by ASMedia eliminating the need for the chipset fan that everyone hates. The chipset on b550 isn't pcie4. Just the lanes directly to the CPU.
|
|
#
¿
Sep 9, 2020 13:08
|
|
|
ConanTheLibrarian posted:One other point: the APUs are monolithic. A chiplet-based APU would be pretty interesting since they could release models with different sized GPUs. Smaller, lower power for laptops versus larger and more capable for desktops. Otoh the benefits of a single die may be too much to pass up for anything bound for mobile form factors. As of right now, idle lower consumption would be a real problem for the chiplet based designs in mobile space. Maybe you can make that better going to a better process for the io die but then you sacrifice some of the cost benefit of doing it in the first place.
|
|
#
¿
Sep 22, 2020 04:10
|
|
|
How good is anyone at keeping anything in stock these days? PSUs, cases people want to buy? Global supply chain is still turbofucked.
|
|
#
¿
Sep 29, 2020 13:05
|
|
|
EmpyreanFlux posted:??? Good for a stock cooler, sure, but still pretty pointless. Unless you just don't care about noise at all or are on a tight budget I'd replace with a tower cooler 100% of the time, to the point where I wish you could buy the cpus without.
|
|
#
¿
Oct 3, 2020 03:52
|
|
|
Klyith posted:It wasn't because they were just going to be replaced, it was that some people did use them and they were totally inadequate. As in the processor would hit max temp and throttle. The sinks they came with were the same cheap intel aluminum sinks as the normal desktop processors, which keep up with a 65W cpu and that's about it. It was making them look bad that people would have awful results, and designing a decent heatsink like the prism would have cut the profit margin. In my perfect world, they'd cut prices on the processors by $5 and not include a heatsink. Then make the max available as a bundle with processor purchase regardless of sku for +10 and the stealth for +5. Boom, less waste for people who are going to immediately replace with something better, more options for people on a tight budget. Probably a pain for logistics and packaging reasons I guess.
|
|
#
¿
Oct 3, 2020 20:16
|
|
|
Ironically AM4s long life is actually a downside rather than a positive as lack of an upgrade path would stop people spending silly amounts of money on marginal upgrades most of them won't notice. Curse you AMD!!!
|
|
#
¿
Oct 4, 2020 19:24
|
|
|
EmpyreanFlux posted:Spires are usually adequate for stock, and the copper cored ones seem to have similar performance to an NH-L9a (per phoronix https://www.phoronix.com/scan.php?page=article&item=noctua-nhl9-am4&num=1). It sucks complete rear end they gutted the Spire recently though. I mean noise is pretty important when evaluating a cooler. More noise for the same result is... bad. Even on a tight budget something like the pure rock or pure rock slim are absolutely worth the upgrade and they are far from the high end (nh-d15s, dark rock pro4) or even midrange (something like the fuma 2).
|
|
#
¿
Oct 4, 2020 23:14
|
|
|
EmpyreanFlux posted:Look at the noise level again in the article, the their lower settings the EVO 212 is 34dB and the Max is 34.8dB. For 150W, Delta over ambient is 31°C for the EVO 212, for the Wraith it's 31.9°C. That's directly comparable. It's only @ full load does the EVO 212 pull ahead in noise in and thermally, but that's @ 40.4dB (to the Max's 44.7dB). Notice the Delta between the lower and max settings though, it's not really worth it to run either at full blast. The other issue is that you're comparing it to the hyper 212, which itself is substantially outperformed by... Other low end tower air coolers. The statement "it's only worth replacing if you go extremely high end" is just not true when there are options between $35-$50 that absolutely dumpster it in acoustic efficiency.
|
|
#
¿
Oct 5, 2020 13:07
|
|
|
Eletriarnation posted:I get what you're saying, but there's no conflict in this. It's a standard recommendation because it's cheap, straightforward to install and has broad compatibility. It won't keep up if you go over around 120W but that's fine for most builds, I used them myself on an i7-920 and a 2500K. When I replaced that i7-920 with a six-core Westmere Xeon and started really pushing it hard, I had to get a big Noctua with six heatpipes because you can't push 200W through a 212. I'm saying it shouldn't be the standard anymore - it's been surpassed by things like the aforementioned pure rock slim at the same price point and similar compatibility. The hyper 212 recommendation is obsolete.
|
|
#
¿
Oct 5, 2020 17:42
|
|
|
Wait ignore everything I just said I got their models crossed - its the shadow rock slim, not the pure rock slim. So the reasonable bequiet options start at $45 for the pure rock non slim which I guess leaves a place for the hyper 212 in cost limited builds.
|
|
#
¿
Oct 5, 2020 17:54
|
|
|
MikeC posted:Expect 0 price drops. These prices are specifically designed to keep 3000 series relevant. That Aussie SI was right on the money. I wouldn't expect them to keep making the 3000 series. They directly compete for fab capacity.
|
|
#
¿
Oct 8, 2020 18:00
|
|
|
Is zen2 cheaper even? At this point r&d is a sunk cost and they're on the same process. I guess we don't know die size but based on core counts and cache I imagine they're pretty close.
|
|
#
¿
Oct 11, 2020 17:49
|
|
|
Nomyth posted:Do we have any indication of how well Zen 3 does for inter-CCX communication for parts > 8 cores? From Epyc examples and such? Or is it the same fabric from zen 2 still Parts over 8 cores would be via the io die, presumably. Since each ccx is now 8 cores, unless they go more than 8 cores per chiplet it seems like inter ccx should be a thing of the past if you're on the same chiplet. Obviously we'll need a deep dive on the uarch to know anything for sure.
|
|
#
¿
Oct 12, 2020 21:18
|
|
|
Pollyanna posted:o You could buy an extension cable pretty cheap and probably find a way to route it from the front port to the rear if you wanted.
|
|
#
¿
Oct 18, 2020 19:24
|
|
|
Tiger lake took the IPC gains of ice lake but actually clocks decently because of 10nm process refinements, so it kind of makes sense that it could beat the 14+++ desktop chips in single threaded benchmarks, particularly if they're on the shorter side and spend longer relative time at the short duration power limit.
|
|
#
¿
Oct 18, 2020 21:16
|
|
|
|
| # ¿ May 22, 2024 08:56 |
|
|
From a pure value proposition, zen3 is going to be worse than zen2 at launch which will probably help make it more available. The entry level sku at 6c 12t is now $300, 8c/16t starts at what, $450? There is no $200 3600, $330 3700x, or $450 3900 this time around.
|
|
#
¿
Oct 21, 2020 03:22
|
|