ECC is already enabled on Ryzen. The could and obviously are validating it anyways. Not only that but it isn't a big issue for general use workstations. It hasn't been in the past with OEM's offering the same workstations or workstation families with i7's that don't have ECC at all. ECC is good thing to have on a workstation, but never has been a deal breaker. So why take a CPU that already has ECC and not validate it, then create another separate platform using the exact same chip and configuration and validate that and then start the cycle all over again with the X399 stuff. Which again brings the other annoyance. Why even offer the full Ryzen lineup or at least down to the 4 core offerings since the Pro space has always wanted more cores anyways and Ryzen's greatest plus has been more competitive cores at a great savings. The 4 core options is the absolute bare minimum and honestly the R5 1400 and 1500 options are selling well enough to eat the actual salvaged dies and therefore any Ryzen Pro offering would just be cannibalizing good cores that could sell for more. Any workstation offering a R5 Pro isn't going to be worried about ecc memory and they could just plop a normal R5 into it. None of this makes any sense.
T1beriu
Member
- Mar 3, 2017
- 165
- 150
- 81
Stupid and pointless. I don't even know why the feel the need to put it on the roadmap. It's almost seems like a vendor specific naming requirement to close a deal.
I'd like to see another FirePro APU, possibly thats what a Ryzen Pro could be, APU wise?
I really wish we got more info on the APU server chip at some point, its been a long time since theres been even a mention of it on a roadmap.
I really wish we got more info on the APU server chip at some point, its been a long time since theres been even a mention of it on a roadmap.
T1beriu
Member
- Mar 3, 2017
- 165
- 150
- 81
No.
Ryzen Pros are Ryzen CPUs sold to big OEMs (HP, Dell etc) to go to commercial business sector.
There are 2 types of Ryzen Pro: one for the desktop - Ryzen Pro (Summit Ridge - just CPU) and one for laptops - Mobile Ryzen Pro (Raven Ridge - CPU + GPU).
Pros come without enthusiastic features because businesses don't need them, like OC or XFR (proof) and come with some bundled software that companies might use to manages the PCs
T1beriu
Member
- Mar 3, 2017
- 165
- 150
- 81
I can't see that being the case because they are very different products marketed for two different workloads. I can't seem to find any FirePro related workloads in the Pro marketing materials.
Pro APUs differentiation with consumer APUs is in security features and a guaranted support for a longer time.
http://www.anandtech.com/show/10727...-bristol-ridge-pro-apus-with-extended-support
Supposedly the Pro Map were supposed to be the best APUs of the bunch. The ones which can't lower the performance on heavy loads.
Right now I have to agree with you. But the Pro line was supposed to be similar to the Pro line of Intel: chips which don't get throttled on heavy works even on the most basic chip since it were the best chips of the bunch.
Now with Ryzen we finally we will see a real differentiation between Pro and Consumer.
T1beriu
Member
- Mar 3, 2017
- 165
- 150
- 81
Where did you get all this from? Can we have a source please?
There's no mention of this in the marketing materials in the Pro lineup from the last 4 years (Richland, Kaveri, Carrizo, Bristol Ridge). If you bring an advantage to the table you would be silly not to advertise it, especially when it's a big selling point like this. "Guys, you don't like throttling chips? Then don't buy the regular one, buy the Pro ones!"
See Abwx's post above what the Pro lineup is for.
Last edited:
AtenRa
Lifer
- Feb 2, 2009
- 14,001
- 3,357
- 136
More details about Raven Ridge.
This 40% higher GPU perf is with 2133MHz DDR-4 ??
Probably they've divided the peak TFLOPs between the 2 and that's it. 8*1.4=11.2, after all, you know.
CatMerc
Golden Member
- Jul 16, 2016
- 1,114
- 1,149
- 136
Yeah, that's my thinking as well. Though it would be a strange way to measure things, considering Vega should provide higher perf per TFLOP. They could have used a higher number there.
Valantar
Golden Member
- Aug 26, 2014
- 1,792
- 508
- 136
You truly seem unable to grasp the point of a 'Pro' lineup. A hint: it is in no way meant as a consumer facing lineup with benefits truly tangible for consumers. 'Pro' doesn't in any way mean 'better performance' or anything remotely similar. What do business users need? Stability, software validation (as in: send samples to makers of leading business software, and work closely with them ironing out any and all bugs found. Which is expensive), long-term support cycles (so you can get a drop-in replacement CPU even 4-5 years down the line if it fails), possibly ECC support (which Ryzen already has), and a robust ecosystem. Performance usually comes after all of these. There will undoubtedly be Pro-branded Threadripper (or low-end Naples) SKUs for business HEDT, but somehow squeezing a 12+c design into AM4 makes no sense - if it could be made to work at all, it would cripple the chip (no 4-channel memory, too little PCIe, so on) for no tangible gain.
Consumer CPUs have 1-2 year product cycles. Business CPUs have 5+-year cycles - and due to OEM agreements, AMD has to produce them for that time. Imagine the silly and wasteful cost if they had to produce every single Ryzen SKU for 5 years because some OEM used them in a few PCs. That's a big part of the reasoning behind the separation (although software validation and recovering the very high cost of that is probably as important).
Just to make it abundantly clear: CPUs for Socket AM4 in the first Zen generation are 8c max. We don't know about Zen+/2/3/etc, but there will not be 12+c SKUs for AM4. That's what X399/Threadripper is for. I'm kind of baffled that you're whining about this in this way. AMD shook up the CPU market just two months ago by bringing >4c designs into the mainstream. Now you're pissed that the mainstream tops out at a "measly" 8c. What are you on about? Ryzen Pro is for professional/business/enterprise markets. It's not meant for consumers at all. Period. They probably won't even be sold at retail. Thus, they won't "confuse the market".
And do you really not understand that businesses and business OEMs want the extra security in the Pro SKUs being developed slower and tested for longer? Do you really think they're dumb enough to be confused by the two? I wouldn't be surprised at all if there was a 'Pro' Ryzen 9 Threadripper lineup, as relegating business HEDT entirely to Naples seems odd, yet they still want/need the same features described above. Some will go for consumer CPUs, sure. But most businesses are conservative, and know that higher hardware costs are negligible if that means more uptime, fewer replacements, and guaranteed spare part availability even years down the line. Downtime = lost revenue, after all.
Gikaseixas
Platinum Member
- Jul 1, 2004
- 2,836
- 218
- 106
You obviously missed where my confusion was and how it was cleared up and why you are too late to this discussion. I still don't think that the Pro line should be used the way it is. But I'll get into it later. I never wanted or expected ThreadRipper/Snowy Owl on AM4 and I don't know where in those quotes you got that from. I am going respond to this and then put you on ignore because this is pretty aggressive for an opening discussion in clearing up something that was already cleared up. I have only put one other person on ignore, so you should feel proud of what you have accomplished.
Yet that is what they are doing anyways. Even if they don't produce 8 (and whatever R3 is on top of that), we know what all of these are. They are the same CPU across the board. They are already at 80% fully functional dies right now. By the end Zen they will probably be up to 85-90%. So any production they have to do from that point on they are just going to be binned chips fully working chips. For businesses they don't even have to create the retail packaging. So all they are going to be doing is changing it in the actual chip packaging in China or Malaysia. Something they could do on fluidly or even on demand, as demand is needed. But you are wrong anyways certain markets like military, security, medical, or POS, require production for more than 2 years, what PC vendors want is platform stability so they quicken their validation cycle. Even then Intel forced on them a 2 year cycle.
Of course I know that AM4 is 8 cores max. I still don't get this part. Where you get that I ever thought ThreadRipper was for AM4. My point and the one you somehow missed even though you quoted like 4 of my posts. My confusion was not knowing that AMD sold a "Pro" line before. Oddly it appears to be a single CPU sold to a single vendor, for a single system. So when the Rumors of a Pro line first came out I assumed that was going to be the brand used for X399/TR. It just made too much sense to me, a large CPU with tons of cores aimed and priced for Precision workstations, render systems, developer suites, video and audio editing, it seemed like the perfect Professional non-server CPU and deserved to be different than Ryzen name wize. Ryzen Pro again seemed perfect. After I realized that ThreadRipper was a sub brand for it and they had already sold a rebranded "pro" cpu, I then understood what the Pro options would be. I still don't agree with the branding but that's just it, it's a branding issue and not configuration issue. As for the CPU's yeah it will add to confusion. When half the OEM's will use Ryzen, half Ryzen Pro, and them sharing model numbers, but not the same clock speed and absence of things like XFR. When getting help in the future we will run into issues trying to figure out exactly what they are running because they will know that they have a Ryzen and maybe 1500 or 1700, but not the pro part and figuring out where the performance sit's is going to be difficult. Not that you will get the chance but we will have to agree to disagree on this portion. To me it's not smart having such similar offers with similar names but still trying to differentiate them.
The problem with all of this is the fact that it is all the same hardware. Not a single different. All the security features. All the validation. ECC. And so on is exactly the same thing between all of them. All AMD is doing it basically forcing a motherboard feature set (like TPM) and Vpro like connectivity on systems offering AMD Pro products. But the thing is the OEM's are responsible for getting the boards designed and produced. They can already do this. They have been doing it. They have been doing it for decades with Intel and AMD never needed to do this in the past when they had a competitive CPU. It really seems like they had to offer an APU "Pro" CPU for HP to get a contract for one of their business systems. From there they felt like it might be good idea to expand it. Me I see it as redundant and unnecessary.
Valantar
Golden Member
- Aug 26, 2014
- 1,792
- 508
- 136
Wow, I did not expect my post to be recieved that harshly. Oh well, I guess there's no point in apologizing for the tone of a post if the person on the receiving end has ignored me. Too bad.
Now, perhaps we might get this thread back on topic?
I came here looking for discussion of Raven Ridge given what we now know from the Financial Analyst Day. Any takers?
IIRC, we have leaks pointing towards a SKU with 11 CUs of graphics - which seems to go over well with the 40% performance increase stated (if the benchmark is 512SP/8CU Carrizo). OTOH, that doesn't tell us anything about clocks - though they might be conservative in mobile to keep thermals and battery life decent. 4c8t at the top is a given, no? My main question is simple: What TDP will this launch at, and how much will it power throttle? Even though the 65W TDP of the R5 1400 is more than it actually uses, I can't imagine powering a similarly clocked 4c8t RR APU alongside a ~1000MHz Vega GPU at 35W or less. Even 45W seems very doubtful IMO without serious power throttling when using the iGPU. Then again, Ryzen is very power efficient at lower clocks, so even dropping a few hundred MHz might help here. Ideally, I would hope they could clock gate cores (or at least core pairs) separately. That way, you could have two cores (four threads) running full bore for a game or other heavy workload, and two slower "background task" cores. That would be ideal.
This always seems to be the key problem with integrated GPUs - that they're not given the power and thermal headroom to shine. I have to say I hope they launch a 65W mobile Raven Ridge APU - or at least an OEM configurable 65W mode. 65W is very possible to dissipate in a small form factor laptop (heck, the Dell XPS 15 has a 45W CPU + a ~50W GPU), while giving both the CPU and GPU room to shine.
For the desktop, I hope they unlock the power draw completely, and rate the chips at 95W or even 125W. Even though no motherboards that I've seen have more than three power phases for the iGPU (and other uncore components), this ought to be sufficient if it draws ~50W. That would open up for both the CPU and GPU running at full bore without power throttling. Which would be absolutely fantastic.
Now, of course, Kaveri and older APUs also came in 95W+ SKUs, but those had a very inefficient CPU architecture that easily used up most of that power headroom by itself. Luckily, Ryzen changes this completely.
Now, perhaps we might get this thread back on topic?
I came here looking for discussion of Raven Ridge given what we now know from the Financial Analyst Day. Any takers?
IIRC, we have leaks pointing towards a SKU with 11 CUs of graphics - which seems to go over well with the 40% performance increase stated (if the benchmark is 512SP/8CU Carrizo). OTOH, that doesn't tell us anything about clocks - though they might be conservative in mobile to keep thermals and battery life decent. 4c8t at the top is a given, no? My main question is simple: What TDP will this launch at, and how much will it power throttle? Even though the 65W TDP of the R5 1400 is more than it actually uses, I can't imagine powering a similarly clocked 4c8t RR APU alongside a ~1000MHz Vega GPU at 35W or less. Even 45W seems very doubtful IMO without serious power throttling when using the iGPU. Then again, Ryzen is very power efficient at lower clocks, so even dropping a few hundred MHz might help here. Ideally, I would hope they could clock gate cores (or at least core pairs) separately. That way, you could have two cores (four threads) running full bore for a game or other heavy workload, and two slower "background task" cores. That would be ideal.
This always seems to be the key problem with integrated GPUs - that they're not given the power and thermal headroom to shine. I have to say I hope they launch a 65W mobile Raven Ridge APU - or at least an OEM configurable 65W mode. 65W is very possible to dissipate in a small form factor laptop (heck, the Dell XPS 15 has a 45W CPU + a ~50W GPU), while giving both the CPU and GPU room to shine.
For the desktop, I hope they unlock the power draw completely, and rate the chips at 95W or even 125W. Even though no motherboards that I've seen have more than three power phases for the iGPU (and other uncore components), this ought to be sufficient if it draws ~50W. That would open up for both the CPU and GPU running at full bore without power throttling. Which would be absolutely fantastic.
Now, of course, Kaveri and older APUs also came in 95W+ SKUs, but those had a very inefficient CPU architecture that easily used up most of that power headroom by itself. Luckily, Ryzen changes this completely.
800 MHz - 11 CUs => 40% more SP Float than 1108 MHz - 8 CUs.
800 MHz - 11 CUs => -70% less DP Float than 1108 MHz - 8 CUs. (Bristol Ridge still king in double float)
800 MHz - 11 CUs => -60% less QP Float than 1108 MHz - 8 CUs. (Probably because of double float).
So, yeah its 40% more perf, but not so much on HPC, yet. 15DD(800 MHz) vs 9784_BR(1108 MHz).
(Secret above is that I'm comparing a mobile to a desktop SKU. ~89% is the mobile to mobile for SP float.(15DD@800 MHz/9874_BR@758 MHz)
800 MHz - 11 CUs => -70% less DP Float than 1108 MHz - 8 CUs. (Bristol Ridge still king in double float)
800 MHz - 11 CUs => -60% less QP Float than 1108 MHz - 8 CUs. (Probably because of double float).
So, yeah its 40% more perf, but not so much on HPC, yet. 15DD(800 MHz) vs 9784_BR(1108 MHz).
(Secret above is that I'm comparing a mobile to a desktop SKU. ~89% is the mobile to mobile for SP float.(15DD@800 MHz/9874_BR@758 MHz)
Don't AMD AM4 motherboards have no power limit? If yes, no doubt Raven Ridge would also have no power limit.
Although, how can one tell whether a motherboard is friendly to integrated GPU overclocking?
Although, how can one tell whether a motherboard is friendly to integrated GPU overclocking?
Valantar
Golden Member
- Aug 26, 2014
- 1,792
- 508
- 136
From my (arguably very limited, mainly with an A8-7600 on an ASRock FM2A88X-ITX+ board) experience, AMD iGPU overclocking is no problem at all, both through software and BIOS.
As for no power limit on the motherboard, isn't that ultimately decided by the CPU? Ryzen chips are all unlocked, but even then the X and non-X SKUs have different power throttle points (I believe something like 95W and 125W, AMD stated this very clearly somewhere, but I can't find it right now - it's somewhere in the main Ryzen thead). I'm not sure if this applies when overclocking, though.
Although the A10 line was/is almost entirely unlocked, I believe even they power throttled when the GPU kicked in. A8s and lower have pretty strict power limits.
DrMrLordX
Lifer
- Apr 27, 2000
- 21,634
- 10,848
- 136
That was only for GV-A1 and KA-V1 Kaveri. Carrizo had a more-sophisticated system of throttling and power usage limitation.
T1beriu
Member
- Mar 3, 2017
- 165
- 150
- 81
For the last 4 years the Pro lineup had the same frequency as the non-Pro CPUs. I expect the same to continue with Ryzen Pro.
I know I said Pro would not have XFR, I made the mistake to link the lack of X CPUs with XFR, even though I knew that wasn't the case. I can't find no reasons why XFR should not be present on a non-X CPU.
Last edited:
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 10K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
-
Facebook
Twitter