Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Grabbed this from Drunkenmaster on OCUK. Some pretty interesting tidbits.
First off we had some rumours about weak IMC due to supposedly low speed. This was followed by a rumour that actually Zen had insane efficiency way above Intel and as such lower speeds actually had same performance as higher on Intel.
http://ranker.sisoftware.net/show_r....1b7c4f9c9&l=en
This is a Zen getting 33.99GB/s out of 2133Mhz memory, which has a theoretical maximum bandwidth of 34.128GB/s... meaning, epic efficiency.
For comparison you have
http://ranker.sisoftware.net/show_r....5b3c0fdcd&l=en
Broadwell-e with 3200Mhz memory achieving 74.97GB/s with max theoretical of 102.4GB/s, so around 75% efficiency.
It's a major issue for some enthusiasts with special wants/needs- the same problem plagues users with Z270-based systems to a lesser degree.
It isn't however an issue for someone who builds a typical system, even for high-end gaming, with one or two GPUs, one M.2, and a few spinning or solid mass storage drives.
This is from http://semiaccurate.com/forums/showpost.php?p=283673&postcount=7443 :
It is lower but it is faster If the example scales to all speeds supported on AM4 .
This. AMD isn't going for the very high end with this system. They have set it up for what most people will want and it has the bandwith for a gpu and a m2 ssd which is what most will have at most. Hell it will run the second gpu for those who want to push the system further.
At least MSI X370 Xpower Gaming Titanium has two M.2 (1 PCIe 3.0 x4, 1 PCIe 2.0 x4) and one U.2 (PCIe 3.0 x4) socket.Almost all of them have only ONE m.2 interface, and, in the case of mobos that have two of them, the second one is always supporting PCIe 2.0 only. It is clear that the mobo manufacturers have had to cut corners across their boards due to this limitation.
Or maybe the memory wasn't 2133Mhz
The IMC isn't AMDs design and its a standard IP PHY.
The links are also broken btw.
Nothing broken here from where are you using the links from my quoted bit or from the SA links ?
From what we have seen of Ryzen thus far (both leaked and official from AMD thus far), it looks like this CPU is likely going to be a big win for AMD and enthusiasts. With that said, two things I've noticed have given me serious second thoughts when it comes to buying it - and they both concern the platform rather than the CPU itself.
First, a minor Achille's heel, seems to be that DDR4 memory speeds supported by Ryzen boards seem to be substantially lower than their Intel counterparts. However, given that such extreme memory speeds often yield negligible performance gains, this is only a small issue.
This brings us to the major Achilles' heel that I am seeing with Ryzen.... I have spent a considerable amount of time reading all of the motherboard documentation I can find regarding almost every X370 mobo that that major manufacturers have announced. (This includes MSI, ASUS, Gigabyte, etc.) I've come to the conclusion that Ryzen seems to have a *major* limitation when it comes to available PCIe 3.0 lanes.
Every single X370 mobo that has been announced appears to be seriously limited in this regard - so much so that, in some cases, multiple PCIe slots only support PCIe 2.0. Almost all of them have only ONE m.2 interface, and, in the case of mobos that have two of them, the second one is always supporting PCIe 2.0 only. It is clear that the mobo manufacturers have had to cut corners across their boards due to this limitation.
This seems almost like a deal breaker level issue to me.
Has anyone else taken note of this? Maybe it is not so much of a major issue?
From what we have seen of Ryzen thus far (both leaked and official from AMD thus far), it looks like this CPU is likely going to be a big win for AMD and enthusiasts. With that said, two things I've noticed have given me serious second thoughts when it comes to buying it - and they both concern the platform rather than the CPU itself.
First, a minor Achille's heel, seems to be that DDR4 memory speeds supported by Ryzen boards seem to be substantially lower than their Intel counterparts. However, given that such extreme memory speeds often yield negligible performance gains, this is only a small issue.
This brings us to the major Achilles' heel that I am seeing with Ryzen.... I have spent a considerable amount of time reading all of the motherboard documentation I can find regarding almost every X370 mobo that that major manufacturers have announced. (This includes MSI, ASUS, Gigabyte, etc.) I've come to the conclusion that Ryzen seems to have a *major* limitation when it comes to available PCIe 3.0 lanes.
Every single X370 mobo that has been announced appears to be seriously limited in this regard - so much so that, in some cases, multiple PCIe slots only support PCIe 2.0. Almost all of them have only ONE m.2 interface, and, in the case of mobos that have two of them, the second one is always supporting PCIe 2.0 only. It is clear that the mobo manufacturers have had to cut corners across their boards due to this limitation.
This seems almost like a deal breaker level issue to me.
Has anyone else taken note of this? Maybe it is not so much of a major issue?
At the prices that AMD boards are at the moment, does it matter if you get a first round board now and eventually upgrade?
I'm seeing prices like 79, 89 and 149 for a premium compared to a stripped down Intel board that's in the 150-200 range.
pci-e lanes is negligible IMO.
most people aren't running dual gpu solutions also if you haven't noticed, sli and x-fire are essentially dead (thanks dx12!)
so there are 8 lanes for nvme raid.
or a raid card or a capture card or whatever card.
also like some intel boards you can use a plx chip (like the one on my board in my sig) to get more lanes.
From what we have seen of Ryzen thus far (both leaked and official from AMD thus far), it looks like this CPU is likely going to be a big win for AMD and enthusiasts. With that said, two things I've noticed have given me serious second thoughts when it comes to buying it - and they both concern the platform rather than the CPU itself.
First, a minor Achille's heel, seems to be that DDR4 memory speeds supported by Ryzen boards seem to be substantially lower than their Intel counterparts. However, given that such extreme memory speeds often yield negligible performance gains, this is only a small issue.
This brings us to the major Achilles' heel that I am seeing with Ryzen.... I have spent a considerable amount of time reading all of the motherboard documentation I can find regarding almost every X370 mobo that that major manufacturers have announced. (This includes MSI, ASUS, Gigabyte, etc.) I've come to the conclusion that Ryzen seems to have a *major* limitation when it comes to available PCIe 3.0 lanes.
Every single X370 mobo that has been announced appears to be seriously limited in this regard - so much so that, in some cases, multiple PCIe slots only support PCIe 2.0. Almost all of them have only ONE m.2 interface, and, in the case of mobos that have two of them, the second one is always supporting PCIe 2.0 only. It is clear that the mobo manufacturers have had to cut corners across their boards due to this limitation.
This seems almost like a deal breaker level issue to me.
Has anyone else taken note of this? Maybe it is not so much of a major issue?
Have no-one ever tested pci-e lane scaling, 8\8x works fine for titan X with no penalty, why should you care about "ohh noes, no 16\16X 16\16X SLI?"
Can you show examples of this? From what I see its rather the other way around. AM4 boards being quite pricy.
My Setup will consist of
16GB (2x8) DDR4 3200
1 Video Card
2 SSD's
Maybe a PCI-e sound card. Depends on how well the on board sound is
And in the future, removing an SSD and getting an M.2 drive.
I don't see any issue with this setup. A lot of hubbub for nothing.
I already bought the RAM. My first G.skill kit.Better pick up those DDR4 sticks soon lol, since Christmas the prices have skyrocketed to some stupid prices.
The number of pci express 3.0 lanes is limited. Motherboard manufacturers can install a PLX chip to remedy that, but it also may introduce latency issues. I suspect this will be resolved next year when Zen gets a refresh. It is NOT an issue that affects a whole lot of people though, so I do not consider it catastrophic. If you have 1 m.2 NVME drive , that is unusual , they tend to be quite expensive. I recently bought a Samsung 850 Pro M.2 NVME drive that was on sale for my Ryzen build, but I hardly think I will be able to get another anytime soon until the prices drop.
Can you show examples of this? From what I see its rather the other way around. AM4 boards being quite pricy.
My point exactlymost of the results fall in to the margin of error zone.
https://www.pugetsystems.com/labs/articles/Titan-X-Performance-PCI-E-3-0-x8-vs-x16-851/
From what we have seen of Ryzen thus far (both leaked and official from AMD thus far), it looks like this CPU is likely going to be a big win for AMD and enthusiasts. With that said, two things I've noticed have given me serious second thoughts when it comes to buying it - and they both concern the platform rather than the CPU itself.
First, a minor Achille's heel, seems to be that DDR4 memory speeds supported by Ryzen boards seem to be substantially lower than their Intel counterparts. However, given that such extreme memory speeds often yield negligible performance gains, this is only a small issue.
This brings us to the major Achilles' heel that I am seeing with Ryzen.... I have spent a considerable amount of time reading all of the motherboard documentation I can find regarding almost every X370 mobo that that major manufacturers have announced. (This includes MSI, ASUS, Gigabyte, etc.) I've come to the conclusion that Ryzen seems to have a *major* limitation when it comes to available PCIe 3.0 lanes.
Every single X370 mobo that has been announced appears to be seriously limited in this regard - so much so that, in some cases, multiple PCIe slots only support PCIe 2.0. Almost all of them have only ONE m.2 interface, and, in the case of mobos that have two of them, the second one is always supporting PCIe 2.0 only. It is clear that the mobo manufacturers have had to cut corners across their boards due to this limitation.
This seems almost like a deal breaker level issue to me.
Has anyone else taken note of this? Maybe it is not so much of a major issue?
It's a major issue for some enthusiasts with special wants/needs- the same problem plagues users with Z270-based systems to a lesser degree.
It isn't however an issue for someone who builds a typical system, even for high-end gaming, with one or two GPUs, one M.2, and a few spinning or solid mass storage drives.
DDR-4 3600 is not fast enough? There have been countless RAM on gaming or productivity benchmarks comparing X to Y speed DDR4 speeds and they all mean piss. RAM has so far surpassed the ability of any real device to utilize it fully.
DDR4 3600MHz is behind the 4133MHz that Intel motherboards can achieve in some cases. Again, straight memory bandwidth is hardly telling of performance. We don't know what impact this will or will not have on the final product. In games you can't say it isn't important because Skylake showed definite improvement over Haswell with DDR4 memory and clock speeds in excess of DDR4 3000MHz. We do not see that with Haswell-E and we might not see that with Ryzen. We just don't know yet. The point is, anything we might think about this is speculation at the moment.
I would rather run slower ram with much tighter timings than much faster ram with very slow and loose timing. Brutal bandwidth is not the same as very strict and tight timings. How fast you can access data is far superior than how much bulk transport of data you can move.
Not quite. Again this comes down to CPU architecture and how it behaves in regard to memory. Intel CPUs have almost always shown virtually no change in performance based on latencies. In contrast, Intel CPUs have often shown better gains through raw bandwidth. This doesn't always translate to application level performance, but this is the tendency Intel's designs tend to have with regard to RAM. For the longest time this was thought to be a product of the chipset based IMC's but didn't change even when the IMC became integral to the CPU. AMD CPU's on the other hand showed remarkable performance increases based on timings more than raw bandwidth. Again this brings me back to the topic of architectural differences and how they relate to actual application performance.
I see absolutely no negative conditions imposed by this. For instance the days of having 40 PCIe lanes are LONG gone. Since SLI and Crossfire is going the way of the dinos and the new thing is really high single card performance. I don't see that as an issue either.
Wrong, we need more PCIe lanes than ever before. With high end configurations using up lanes for storage, networking and audio, I don't think the need for 40 PCIe lanes is going to change anytime soon. While its true that SLI and Crossfire may fizzle out, we can't know that for certain. NVIDIA and AMD will do all they can to fight that because it's just good business. Consoles are what hurt us here, but again the PC market shows signs of growth. At least this is what many motherboard and graphics cards manufacturers' research shows. Believe me, I've sat in many presentations from Intel and other companies showing the same data trends. PC gaming is on the rise. If AMD and NVIDIA can sell guys like me two GPUs instead of one, they will. Dual GPUs may not go anywhere in high end systems. It depends on how much influence they can exert on the game developers. Oddly, what hurts us here is DirectX 12 placing multiGPU performance on the devs and not NVIDIA/AMD.
The only issue I can forsee with this new platform is the fact that I have to wait many many more moons before I can get a respectable mITX board for it.
Ryzen supports DDR4-3600 ... I'd say thats almost as fast as you can get without some serious overclocking. Corsair makes one module that sits at 3866. I am unsure of where you are going with this?
I want to see the days of Ram becoming fast enough that GPUs no longer have onboard ram and you can just allocate 128GB of graphics ram in a 256GB system.
I have no idea on the mITX boards for Ryzen. It remains to be seen how well the industry will take to this. We won't really know if its well suited to the form factor until we get processors in hand. We need to see temperatures, power consumption etc. to really know if mITX makes sense for Ryzen. I'm sure ASRock and a few others will make mITX motherboards for it but I don't know how common they'll be. As for RAM, there are 4133MHz modules that exist now. Dedicated GPU memory isn't going anywhere either. System RAM is still far from being on the level as GDDR5 or HBM2.
My point exactly
And strength to the point:
https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080_PCI_Express_Scaling/24.html
pci-e lanes isn't that important as long as it's the last gen that usually last quite a few years before it has any effect and where HEDT owners would have purchased something new to support their 1000 $ + graphics card(s)
Proof to claims ?Enough to make Intel look slower than AMD in their test so AMD can brag about 4% better performance. This is something Intel will contact independent reviewers about it to make sure they don't disable quad memory channel and also do not make 2x16 to run as 2x8 and that's what AMD did.
Who cares if they disable that stuff as long as they make it known in the review under their test config. In and of itself, while somewhat lame, is not that huge of a deal. Covering it up would be a legit issue IMO.
Sometimes it makes sense from a scientific stand point to "level the playing field" in certain ways by clocking things down or up to make them as "equal" as possible.
From a buyer standpoint, the majority of us, we would prefer they don't do that type of thing. It can yield interesting info however as a matter of curiousity.
Just go on Newegg and search 2011v3 motherboards and you'll see a broad range with upwards to 97 boards in the 200-300 (avg) range versus Ryzen only 21 boards currently but the majority is within the 100-200 range even for x370 boards with the top end going for around $299 (ROG and Fatality boards) which is still cheaper than the 2011v3 boards premium boards going for 300-600.
I'm looking at 2011v3 because the current matchup is the 6900K vs the 1800X (X370 / B350).
As far as the 79-149 range I mentioned, the B350 seems capable of handling the 1800X (reserving my opinions until the NDA lifts on OC results). So right now it seems that the only difference with the X370 vs B350 is USB and PCIe lanes and there are a few premium ATX options in the 150 range.
This one is the one I've been looking at for the 1800X
https://www.newegg.com/Product/Prod...44019&cm_re=ryzen_matx-_-13-144-019-_-Product
I dunno at the current rate I see the AM4 boards averaging at the 150-200 range maybe a few ATX boards going up to 300 but def not like the 2011v3 trends atm.
Comparing apples to apples, I think the Z270 and X370 are far closer than X99 to X370 is. Even then, I think Z270 is better on paper at the very least. I think the issue is that AMD targeted Haswell and it's platforms which is why the X370 motherboard platform is so underwhelming.