|
Dramicus posted:3000 officially don't have support for a320, b350, and x370 but they are running on those platforms anyway. I'll wait for board partners to confirm their products won't support it. Thats all well and good for you, but for people needing to make buying decisions between now and whenever Zen3 hits the streets, the advice is gonna have to be get X570 or B550 if you want to be able to upgrade.
|
#
?
May 7, 2020 19:29
|
|
|
|
| # ? Jun 5, 2024 19:10 |
|
|
Cygni posted:Thats all well and good for you, but for people needing to make buying decisions between now and whenever Zen3 hits the streets, the advice is gonna have to be get X570 or B550 if you want to be able to upgrade. Yes, I agree. This situation is complete garbage from a marketing standpoint.
|
|
#
?
May 7, 2020 19:32
|
|
|
I mean, this is incredibly poorly timed considering Rocket Lake is coming up soon and you're basically giving up an advantage by requiring CPU+Mobo instead of just CPU as an upgrade path. I think AMD should have said they can't guarantee Zen3 support due to BIOS size issues across every motherboard and it'd be entirely up to vendors, but a blanket ban just sours me entirely and if Rocket Lake is better I'll probably hop back over to Intel TBH.
|
|
#
?
May 7, 2020 20:00
|
|
|
EmpyreanFlux posted:I mean, this is incredibly poorly timed considering Rocket Lake is coming up soon and you're basically giving up an advantage by requiring CPU+Mobo instead of just CPU as an upgrade path. I think AMD should have said they can't guarantee Zen3 support due to BIOS size issues across every motherboard and it'd be entirely up to vendors, but a blanket ban just sours me entirely and if Rocket Lake is better I'll probably hop back over to Intel TBH. I doubt Rocket Lake will be better, which is why AMD can do this.
|
|
#
?
May 7, 2020 20:05
|
|
|
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
|
|
|
an actual dog posted:I doubt Rocket Lake will be better, which is why AMD can do this. Rocket Lake is "supposed to be" a backport of Willow Cove onto 14nm. Like potential power issues aside from trying to push even fatter cores on 14nm, I still think that's going to push Intel fairly far ahead on anything latency sensitive as well as equal on productivity. I mean Zen3 is only rumored for a 10-15% increase in performance which at best places it equal to a Willow Cove core if not behind*. *My math is Zen2 is 7% faster than Skylake, Icelake is 18% faster than Skylake, Zen3 is theoretically 23% faster than Skylake, Tigerlake is theoretically 17% faster than Icelake and thus 38% faster than Skylake and thus 12% faster than Zen3
|
|
#
?
May 7, 2020 20:13
|
|
|
oh god they did the AM3/AM3+ quagmire again 
|
|
#
?
May 7, 2020 20:17
|
|
|
I think before we make up numbers for Intel's hypothetical lead with Rocket Lake we should let them actually launch and sell Comet Lake first.
|
|
#
?
May 7, 2020 20:18
|
|
|
EmpyreanFlux posted:Rocket Lake is "supposed to be" a backport of Willow Cove onto 14nm. Like potential power issues aside from trying to push even fatter cores on 14nm, I still think that's going to push Intel fairly far ahead on anything latency sensitive as well as equal on productivity. I mean Zen3 is only rumored for a 10-15% increase in performance which at best places it equal to a Willow Cove core if not behind*. I'm guessing that it trades a big improvement in IPC (Sunny Cove/Ice Lake alone is an 18% improvement in IPC over Skylake, this is another generation farther ahead) for a return to Haswell-esque clock rates (due the the architecture not really being designed around the less dense 14nm node and thus suffering loosened timings). The benefit of the timings being all fucky and reducing clocks is that they won't be pumping voltage to the gills. I actually think it will net out to be the opposite, the fat core improves IPC and the processor nets out to be more efficient/cooler, while being moderately faster (say, something in the 10-15% range). I mean, new uarchs are kinda supposed to be better, Intel just hasn't had one they could deploy on 14nm. If it does clock to 5 GHz or higher, that's a different story, power won't drop that much, but 5 GHz with >>> 18% IPC gain would be pretty tasty too. Paul MaudDib fucked around with this message at 20:26 on May 7, 2020 |
|
#
?
May 7, 2020 20:20
|
|
|
It's almost like there's no such thing as a "good guy" corporation, everything is a business decision, it all comes down to the bottom line, and "brand loyalty" a lie marketing departments made up to get you to make decisions that align with their interests rather than your own. This is a business decision, pure and simple. You can tell this because (see above) they're all business decisions. It's what AMD calculates will generate the most profit for themselves and their partners. AMD's goal is the opposite of remaining the "best buy" or "bargain option". They want to move upmarket and get those margins nice and fat. There will be profit taking now that they're not forced into the low end. The best we can hope for, as consumers who benefit from competition in this market, is that AMD uses those increased margins and sales to fund R&D so that they don't just get clobbered and almost die again when Intel inevitably gets their poo poo back together. We should also make business decisions, and buy the tech that best fits our use case, at the best price.
|
|
#
?
May 7, 2020 20:29
|
|
|
This probably helps Zen2’s value hold up, since if you want to buy a zen3 you have to buy from a comparably limited set of boards.
|
|
#
?
May 7, 2020 20:54
|
|
|
Market dominance has something to do with it, but I also think people underestimate the difficulty or headache it takes to maintain compatibility through different generations. Not helping things are actual real technical reasons that people dismiss.
|
|
#
?
May 7, 2020 21:08
|
|
|
Someone made the point elsewhere that this might be AMD getting their legal ducks in a row after the last Zen 2 issue with advertized clock speeds. Better to be too categorical than saying "No, asterisk".
|
|
#
?
May 7, 2020 21:18
|
|
|
ufarn posted:Someone made the point elsewhere that this might be AMD getting their legal ducks in a row after the last Zen 2 issue with advertized clock speeds. Better to be too categorical than saying "No, asterisk". It certainly would solve a lot of potential issues by being able to respond to complaints with "Running the cpu on that platform is unsupported, not our problem."
|
|
#
?
May 7, 2020 21:29
|
|
|
Dramicus posted:It certainly would solve a lot of potential issues by being able to respond to complaints with "Running the cpu on that platform is unsupported, not our problem." Maybe from a legal point of view, but maybe the company doesn't also want to have to deal with the backlash of a very vocal PC Master Race dragging them online over stuff not working. Or motherboard manufacturers coming to them to look at bugs with corner cases they never bothered to fully qual or validation themselves because they made some internal decision at some point that it wasn't going to be on the roadmap. Maybe they just don't want to have to deal with a sea of uncertainty or devote a non-zero amount resources to the task, especially now that people actually want to buy their stuff so they don't have to toss in extras to make them sound more enticing. Maintaining compatibility is not a free thing from a company's point of view WhyteRyce fucked around with this message at 21:40 on May 7, 2020 |
|
#
?
May 7, 2020 21:35
|
|
|
The official excuse is "abloo bloo BIOS ROM not large enough for future CPUs" though, not anything you describe.
|
|
#
?
May 7, 2020 21:56
|
|
|
orcane posted:The official excuse is "abloo bloo BIOS ROM not large enough for future CPUs" though, not anything you describe. This comes after MSI made special "Max" versions of b450 and x470 boards and expanded their bios roms with the express purpose of making sure they would have more room in the future.
|
|
#
?
May 7, 2020 22:07
|
|
|
Some people say they're doing this for the board partners who cheaped out on BIOS ROMs (again) and would have to throw out older CPUs. But I question the wisdom of making yourself look bad (again) in case some board partners hosed up. Technical reasons would be the easier, more credible excuse so I'm assuming they would have used it if it was true (like Intel's move to LGA 1151v2). And I understand the incentive to increase their margins and get people to buy new mainboards, but in this case that's almost entirely on mainboard manufacturers (for AMD's bottom line it would probably be better if people with 300/400 series chipsets could buy a new Ryzen 4000 as a drop-in upgrade) and "make people buy new stuff" also comes with the risk they won't be buying the new AMD CPU at all. So if the manufacturers wanted to compete with their own new boards by adding Ryzen 4000 support to older boards, it doesn't make sense for AMD to step in and tell them they're not allowed to do that. We'll see. It definitely means people who want to keep at least a limited guaranteed upgrade path right now should wait for B550 mainboards, and consider B450 a budget option only if you know you'll never want more than a 3950X (support for anything newer would just be a nice bonus).
|
|
#
?
May 7, 2020 22:35
|
|
|
orcane posted:The official excuse is "abloo bloo BIOS ROM not large enough for future CPUs" though, not anything you describe. I mean, is that incorrect? If AMD knew that there were a mix of boards in the wild that would have compatibility issues, they had to decide between how to control the messaging and fallout or just cutting the whole thing off. I'm sure there significantly better position has something to do with the end solution, but it's an actual valid technical reason. orcane posted:Some people say they're doing this for the board partners who cheaped out on BIOS ROMs (again) and would have to throw out older CPUs. But I question the wisdom of making yourself look bad (again) in case some board partners hosed up. This area I'm not too clear on, but I assume AMD is supposed to provide some kind of reference design kit to the motherboard makers which details things like this. If AMD did not give explicit direction on the type of chip to use, then that's on AMD. quote:So if the manufacturers wanted to compete with their own new boards by adding Ryzen 4000 support to older boards, it doesn't make sense for AMD to step in and tell them they're not allowed to do that. eh, I highly doubt AMD letting the Ryzen 4000 launch be rife with motherboard incompatibility complaints and inconsistent messaging is something they wouldn't care about. You want the launch to go clean and smooth and letting it be hijacked by issues outside of your direct control is generally not part of the playbook WhyteRyce fucked around with this message at 22:55 on May 7, 2020 |
|
#
?
May 7, 2020 22:48
|
|
|
Dramicus posted:This comes after MSI made special "Max" versions of b450 and x470 boards and expanded their bios roms with the express purpose of making sure they would have more room in the future. the MSI Max mobos don't have bigger bios eproms than other motherboards. MSI just has the combo of an elaborate bios GUI with many images, plus being cheap and using 16MB chips on their non-Max boards. gigabyte has 16mb chips on their b450 and x470 boards and had no problems supporting the same CPUs, because they have a way more simple UI.
|
|
#
?
May 7, 2020 22:49
|
|
|
I mean a lot of partners doubled up on the BIOS ROM size with the 400 series, maybe not all but enough that you could buy a board expressly because of that. Seriously, I don't see a reason why AMD would so vocally shoot themselves in the foot. Just say you cannot officially support Zen 3 on older platforms but that older platform support is dependent on OEM partners. I mean poo poo there is next to no danger running Zen2 silicon on the worst of A320 boards, the CPU will just go "Wow, crazy, power delivery sucks rear end, guess I'll automatically throttle to the best the board can support!" I don't see the design limitations on older AM4 boards that won't prevent BIOS editing for support of Zen 3 anyway, I think it'll happen it just won't be officially supported in any capacity.
|
|
#
?
May 7, 2020 22:52
|
|
|
Klyith posted:the MSI Max mobos don't have bigger bios eproms than other motherboards. MSI just has the combo of an elaborate bios GUI with many images, plus being cheap and using 16MB chips on their non-Max boards. MSI posted:Our MAX motherboards variants come equipped with a 32 MB BIOS chip (instead of a 16 MB chip on older B450 motherboards) that allows for support for all AM4 processors supported by the chipset and support for the familiar, full-featured MSI UEFI BIOS (Click BIOS 5) https://www.msi.com/blog/msis-max-motherboard-lineup
|
|
#
?
May 7, 2020 22:58
|
|
|
Also calling out or singling out partners for breaking Zen 3 compatibility (assuming AMD was explicit in what they needed to do in order to have it and they ignored it) is probably not a smart business move either
|
|
#
?
May 7, 2020 22:59
|
|
|
Klyith posted:the MSI Max mobos don't have bigger bios eproms than other motherboards. MSI just has the combo of an elaborate bios GUI with many images, plus being cheap and using 16MB chips on their non-Max boards.
|
|
#
?
May 7, 2020 23:07
|
|
|
I guess they need to use 10 mb for that sick dragon picture.
|
|
#
?
May 7, 2020 23:09
|
|
|
yes? that's what I said. 32mb is a normal size with other mobos on the market, it's not special. MSI just hosed up. (and then turned it into a giant advantage by having a board that people could order online knowing it would work OOTB.) edit: I have a MSI x370, and I'll say this in their favor: some of the GUI extras are actually pretty neat. like this thing, which you can hover your mouse over to see what's connected and in use. If I can't remember if SYS_FAN3 is the one in the lower right or center left, I don't have to dig out the manual. And I really like the fan curve GUI. So the over-size bios isn't all just dumb pics. But they should have just used 32mb chips in the first place, they knew their bios is fat. vvv e: gotcha, guess "other motherboards" is not clear what other means Klyith fucked around with this message at 23:20 on May 7, 2020 |
|
#
?
May 7, 2020 23:10
|
|
|
Klyith posted:yes? that's what I said. Sorry, I thought you said they didn't increase the size of the rom. I thought "don't have bigger = they aren't bigger than before", my mistake.
|
|
#
?
May 7, 2020 23:14
|
|
|
EmpyreanFlux posted:I mean a lot of partners doubled up on the BIOS ROM size with the 400 series, maybe not all but enough that you could buy a board expressly because of that. Seriously, I don't see a reason why AMD would so vocally shoot themselves in the foot. Just say you cannot officially support Zen 3 on older platforms but that older platform support is dependent on OEM partners. I mean poo poo there is next to no danger running Zen2 silicon on the worst of A320 boards, the CPU will just go "Wow, crazy, power delivery sucks rear end, guess I'll automatically throttle to the best the board can support!" AMD didn't even want to have Zen2 running on 300-series boards in the first place. This time they are putting their foot down. https://twitter.com/hardwareunboxed/status/1258411872539176960?s=21
|
|
#
?
May 7, 2020 23:42
|
|
|
Whether AMD had technical reasons or not for the decision is kinda irrelevant to the people who made purchasing decisions partly due to AMD repeatedly promising an upgrade pathway through 2020. It also makes all the snide AMD marketing slides about platforms look even dumber than AMD's normal dumb marketing slides. I believe they even called Intel's 2 year platform cadence a "scam" on one, and now have basically established their own 2 year cadence in an even shittier and more confusing way. I believe B450 owners specifically have every right to be pissed, considering they didn't even have the option of buying B550 when they made their decision due to AMD delaying the product nearly a year. I feel bad for the people i built/recommended 3600+B450 builds to the last few months. Zen2 has been a hell of a disaster of a launch for such a fantastic performing product in a vacuum.
|
|
#
?
May 7, 2020 23:43
|
|
|
Cygni posted:I believe B450 owners specifically have every right to be pissed, considering they didn't even have the option of buying B550 when they made their decision due to AMD delaying the product nearly a year. Yeah this is what makes me think the HWunboxed guys are getting a story that may not be the final decision, or it's AMD being over precious about the definition of support again. If they'd had B550 available for the entire last year along with X570, so that people weren't buying 400 series mobos this whole time, AMD might be able to drop support for the 400s and get away without people being too mad. If this is the real deal it'll be a giant PR loss when they've been winning all those battles until now. Intel won't have hosed up fabs forever, banking goodwill is smart.
|
|
#
?
May 8, 2020 00:02
|
|
|
Yeah until recently the claim was "AM4 support until 2020" with emphasis on missing out on new chipset/mainboard features (but lack of basic compatibility was never really noted). Even if it was for a technical reason, they had plenty of time to notice and point out that the common assumption (that current mainboards would be able to use the next CPU, at the cost of advanced mainboard features), did not apply beyond Ryzen 3000. Now they have "socket compatibility" as a marketing feature but some of their boards have exactly two CPU generations worth of support, same as Intel.
|
|
#
?
May 8, 2020 00:03
|
|
|
A lot of people probably bought into Zen 2 after holding off to see how Zen and Zen+ would play out. After seeing the compatibility claim be supported twice, and with no b550 available. They probably bought into b450 with the understanding that it would be compatible for at least another release. I might seriously consider going back to Intel if AMD doesn't keep their word and do the right thing in this case. Again, it's probably too early to tell yet and the partners have yet to say anything. But if AMD truly is "putting their foot down" then gg.
|
|
#
?
May 8, 2020 00:15
|
|
|
people are already calling the 500 series "one chip wonders" lmao
|
|
#
?
May 8, 2020 01:28
|
|
|
Doing the mental run-through of an upgrade path for all CPU gens even if you're size limited doesn't seem to make this that complicated. I don't know what's stored in BIOS for each supported CPU gen but assuming it's roughly the same size for each gen and that some significant subset of BIOSes cannot store more than 3 times that size of CPU-specific data. Call a BIOS config that supports gens 1/2/3 a 123 BIOS. Then you would need the following BIOS permutations to allow upgrading from a 123 bios to a gen 4 chip: 134 234 Ie., say you're on a gen 1 chip and want to go to a gen 4 chip. You upgrade to the 134 BIOS and then plop in the gen 4 chip. After which you can update to the latest and greatest 234 BIOS going forward. The 234 will be in the B550/X570, so that's one whole extra BIOS download, 134, you have to offer before folks can upgrade to a new chip. The real concern is going to be users shooting themselves in the foot I guess. E: Put another way, everyone that's not on a gen 1 chip just updates to the 234 bios which is the default bios and moves on with life. If you're on a gen 1 chip you have to use a 134 bios as a special one time upgrade bios then go to the normal 234 bios like everyone else. Only the gen 1 folks have to do anything special. Lots and lots of assumptions, but I still think AMD is being unnecessarily silly here. v1ld fucked around with this message at 01:58 on May 8, 2020 |
|
#
?
May 8, 2020 01:41
|
|
|
Guess the reference RX480 slot burnout or the 5600XT 12/14GBps fiascos taught AMD nothing about to be more careful about product launches.
|
|
#
?
May 8, 2020 01:41
|
|
|
Something that got missed in the R3 launch: https://www.youtube.com/watch?v=xDmRpcQQPUo Holy poo poo, AMD has zero disregard for Intel, the 4700U here has an amazing spec sheet. If I were inclined to crime, I might have mugged someone for this back in college. SwissArmyDruid fucked around with this message at 04:23 on May 8, 2020 |
|
#
?
May 8, 2020 03:15
|
|
|
SwissArmyDruid posted:Something that got missed in the R3 launch: https://www.youtube.com/watch?v=xDmRpcQQPUo I think that's the advantage of Renoir really, not even being a performance leader but AMD being willing to stick something like a 4700U into budget laptops compared to Intel gatekeeping the noticeably better Icelake CPUs. For consumers this matters because it means AMD will be going up against Intels budget laptops CPU, which will absolutely crush what Intel can offer. Like it's been noted AMD has had to offer noticeably better performance to get marketshare, this is it I think.
|
|
#
?
May 8, 2020 05:17
|
|
|
wait, so is the TL:DR of this no Ryzen 4800X (or whatever it ends up being called) on any X470/B450 board?
|
|
#
?
May 8, 2020 05:25
|
|
|
Mr Chips posted:wait, so is the TL:DR of this no Ryzen 4800X (or whatever it ends up being called) on any X470/B450 board? Until we hear from board partners, we have to assume that's the case for now.
|
|
#
?
May 8, 2020 05:29
|
|
|
|
| # ? Jun 5, 2024 19:10 |
|
|
I'll honestly be surprised if it turns out to be a blanket X470/B450 = No! I do personally feel a bit relieved that I decided to spring for an X570 on the build I just completed, as I had been considering going with an X470 but decided to go with the newer board to try buy a little more future-proofing.
|
|
#
?
May 8, 2020 06:17
|
|
