News AMD's next big launch is set on October!!

JaSoN_cRuZe

Honorable
Mar 5, 2017
457
41
10,890
According to some latest info, AMD is expected to launch the Navi 12 (RX 5800) and the Navi 14(RX 5600) graphics cards next month along with the Ryzen 5 3500 and the 3500X and the B550 chipset-based motherboards.
 
Well, the requisite 3 month waiting period has elapsed between X and B board releases.

I just hope the rumor about the B550 having PCIe4.0 on the primary M.2 slot is true. That's all most people want/need..... and no chipset fan.
 
Last edited:

JaSoN_cRuZe

Honorable
Mar 5, 2017
457
41
10,890
Are they suggesting, by the pic in the article, that MSI will put a 'Godlike' class B550 board on offer? All those extreme overclocking features shown would make AM4 motherboard board product stacks even more confusing and difficult to understand where the best values are to be found.

I personally believe that there will some major improvements in the VRM category when compared to x470 & B450 motherboards, as we have already seen the overkill VRM components from the x570 series.

Apart from that they will be priced significantly lower because of the restriction of using only PCIE3 lanes instead of PCIE4 of x570 , Cost effectiveness and VRM improvements without the horrible chipset fan is a must otherwise there won't be a need to get a B550.
 
I personally believe that there will some major improvements in the VRM category when compared to x470 & B450 motherboards, as we have already seen the overkill VRM components from the x570 series.

Apart from that they will be priced significantly lower because of the restriction of using only PCIE3 lanes instead of PCIE4 of x570 , Cost effectiveness and VRM improvements without the horrible chipset fan is a must otherwise there won't be a need to get a B550.

16 core CPU's are still waiting to be seen, but even the current B450's have been shown very capable of running Zen2 processors reliably at reasonable operating points. You don't have to give EVERY board a VRM capable of holding 16 core processors at 5Ghz overclocks under LN2.

I think the problem is more of churn: the board partners looking for a way to force upgrade to new boards as Intel did. All you have to do is look at how MSI is slow-rolling BIOS updates and gimping their own B450 board line for it.
 

JaSoN_cRuZe

Honorable
Mar 5, 2017
457
41
10,890
16 core CPU's are still waiting to be seen, but even the current B450's have been shown very capable of running Zen2 processors reliably at reasonable operating points. You don't have to give EVERY board a VRM capable of holding 16 core processors at 5Ghz overclocks under LN2.

I think the problem is more of churn: the board partners looking for a way to force upgrade to new boards as Intel did. All you have to do is look at how MSI is slow-rolling BIOS updates and gimping their own B450 board line for it.

The problem with embracing the current B450 over the upcoming B550 is just because of the BIOS, Everything except MSI have terrible VRM and bad heat sinks over it. The problem with MSI is that they have screwed the BIOS so much that users are experiencing weird boot issues and the cherry on top is the lower capacity BIOS chip which makes the BIOS look like a retro movie. They tried to provide MAX versions with bigger BIOS chips just for selected motherboards to preserve their x570 sales margin.

So the bottom line is waiting for B550 boards is just an expectation to see better featured boards with better VRM's altogether and from different brands at affordable price.
 

InvalidError

Titan
Moderator
I just hope the rumor about the B550 having PCIe4.0 on the primary M.2 slot is true. That's all most people want/need..... and no chipset fan.
The chipset has no PCIe 4.0 support, so whether the M2 slot can have PCIe 4 is contingent on AMD deciding not to arbitrarily disable PCIe 4 on the CPU's PCIe lanes when paired with a B550 chipset.

I really hate this CPU manufacturer arbitrarily crippling their CPUs based on what chipset it is connected to BS. I was entertaining a slight possibility of upgrading to a 3600/B550 but if AMD is indeed crippling CPUs based on chipset, I'm skipping to 4th-gen or later, crossing fingers that the stuff that got aribitrarily crippled in 3rd-gen will become baseline.
 
...The problem with MSI is that they have screwed the BIOS so much that users are experiencing weird boot issues and the cherry on top is the lower capacity ...
Luckily the internet has stepped in to save us MSI B450/X470 owners... namely, The Stilt has been patching legacy BIOS's to bring them to current AMD specs without breaking them. I'm running an extremely stable patched bios of his on my B450M mortar with a 3700X. Regular and frequent boosts to 4.4Ghz in games.

I'm very happy with the new clik BIOS...it's fast, easy to navigate, intuitive. Some of the features (like Overclocking Profile save/load) could be better located in the menus, but that's a one-time thing once you learn it. I'm looking forward to rarely having to enter BIOS once set up is stable, so wasting time and resources on a game-looking high resolution screens is an utter waste.
 
The chipset has no PCIe 4.0 support, so whether the M2 slot can have PCIe 4 is contingent on AMD deciding not to arbitrarily disable PCIe 4 on the CPU's PCIe lanes when paired with a B550 chipset.

I really hate this CPU manufacturer arbitrarily crippling their CPUs based on what chipset it is connected to BS. I was entertaining a slight possibility of upgrading to a 3600/B550 but if AMD is indeed crippling CPUs based on chipset, I'm skipping to 4th-gen or later, crossing fingers that the stuff that got aribitrarily crippled in 3rd-gen will become baseline.
So far at least, PCIe gen 4 on one M.2 is only a boon to synthetic sequential read benchmarks. Real-world in the OS is frequent random reads where things slow way down. So unless your life is spent copying huge files around it seems a bit pointless to me.

I'm waiting for a use case that truly benefits me as an average user before I get all giggity over PCie gen 4.
 

InvalidError

Titan
Moderator
I'm waiting for a use case that truly benefits me as an average user before I get all giggity over PCie gen 4.
There is a very simple use-case scenario for PCIe 4 SSD right now: swapfile. If you ever run low on RAM, a swapfile on a fast PCIe 4.0 SSD is going to perform much better than a 3.0 one. While 16GB of RAM may be enough for most everyday uses and is quite reasonably cheap at the moment, most people likely have days where they could use more, or at least a faster SSD to more smoothly sail through those times.

You may not need it today but if you keep your PCs for 5+ years like a growing number of people do (still using an i5-3470 as my main PC right now and a C2D E8400 in my living room), you will very likely need it within the computer's useful life.
 
  • Like
Reactions: CountMike
....
You may not need it today but if you keep your PCs for 5+ years like a growing number of people do (still using an i5-3470 as my main PC right now and a C2D E8400 in my living room), you will very likely need it within the computer's useful life.
I don't find a swapfile use case all that compelling, with what I've been doing at least the 16Gb memory I have proves more than enough and I'd rather upgrade to 32Gb if I did find such a need. That's got to be cheaper than an X570 mATX and PCIe gen 4 NVME.

But it's that old argument of buying for the 'future proof' that always makes me a bit nervous. In 5 years everything will change; even Zen5 will be but a memory. I really want to buy for now and use it up for what it is. If the next thing is that beneficial I'll consider it as soon as budget allows but do it right at the time. I've been around long enough to observe the next thing being a flop more often than a flip. But that's all philosophical and opinion which, since everyone has their own upgrade policy and budget constraints, naturally has to be a very personal thing
 

InvalidError

Titan
Moderator
But it's that old argument of buying for the 'future proof' that always makes me a bit nervous. In 5 years everything will change; even Zen5 will be but a memory.
Five years from now, Zen 5 will be less than two years old, possibly still in volume production. Most people will still be perfectly happy with their Zen 2 and 8th-10th-gen Core systems, much fewer will be anywhere close to contemplating upgrading from Zen 5 yet. Forums today still regularly see people only now looking to upgrade from Nehalem-Haswell and AM3/FM2 era PCs, which seems like a clear indication there is a significant number of people already holding on to their PCs for ~10 years today, not going to get any better five years from now as progress on CPU performance per dollar is still abysmal compared to what it used to be.

If your future-proofing horizon is only 2-3 years, sure, upgrade based on what you need this week every week you need something extra. If you expect to still be using the system in some meaningful capacity 5+ years down the road though, future-proofing at least a little beyond the immediate future can go a long way. Seven years ago, SATA3 (6Gbps) was considered useless since no HDD could come anywhere close to SATA2's peak bandwidth and what few SSDs were available were stupidly expensive. Today, the ability to hook up spacious, fast yet affordable SATA3 SSDs is the saving grace of many of those systems. I can certainly appreciate Windows updates taking ~20min instead of 2-3 hours when I do them every couple of months on my PC.
 
Will there be a use for PCI-E 4.0 bandwidth for the average user in the future? Maybe, maybe not, but most people don't know what a swap file is let alone it's existence. I'm waiting for the rumored 10 core Comet lake cpus for desktops in 2020. I'm all set with purchasing a motherboard with a chipset fan and i'm certainly not going to spend $700 for a x570 motherboard that's passively cooled. I'll stick with PCI-E 3.0 for Comet lake and demote my Ryzen R7 1700 from gaming machine to video editing so it can sit there and render in the background. Until then I'll stick with what I have and agree that you only need so much processing power. For me personally, purchasing internal system components now would be a poor decision. I'd like to see Intel's response to third gen Ryzen competition.
 
Last edited:

InvalidError

Titan
Moderator
Will there be a use for PCI-E 4.0 bandwidth for the average user in the future?
Did anyone expect to have a use for SATA3 in the future back when SATA3 was new? Not really.

A few years from now, cheap M.2-4.0 SSDs will be faster than the best available today and an upgrade to M.2-4.0 storage will be well worth it to those who have a board capable of supporting it and can be bothered with the upgrade. I do a clean Windows re-install on my PC every 3-4 years and whenever I do so, it is on a new drive so I can keep my previous install intact until I'm done re-installing everything and copying my old data over. In my case, this would likely translate to building my next system with a SATA3 SSD initially and upgrading to M.2-4.0 when I re-install Windows on it 3-4 years down the line with the expectation that NVMe M.2 will be cheaper than SATA by then due to volume, M.2 using a card edge connector (free) and not having a housing.
 
The benefits of having a swapfile on a faster SSD seem a bit questionable to me. You'll still be taking a performance hit either way when accessing data that has been offloaded to the swapfile, so it would likely be better to have more RAM if optimal performance is a concern.

Even the benefits of having an SSD that maximizes the throughput of PCIe 3.0 still seem a bit tenuous. Certainly there are use-cases involving large reads and writes where that extra throughput could help, but the vast majority of users currently won't see much difference in real-world performance between a $200+ 1TB SSD, and one that costs half as much for the same capacity but lower sequential performance.

Seven years ago, SATA3 (6Gbps) was considered useless since no HDD could come anywhere close to SATA2's peak bandwidth and what few SSDs were available were stupidly expensive. Today, the ability to hook up spacious, fast yet affordable SATA3 SSDs is the saving grace of many of those systems.
Except you can still get most of the benefits of a SATA 3 SSD on a SATA 2 port. Sure, sequential performance will be limited to under 300MB/s, but most of the performance benefits of an SSD over a hard drive come down to low access times improving the performance of random reads and writes, not sequential performance, and SATA 2 will still handle that pretty well.

As a counterpoint, how about PCIe 3.0 vs 2.0? 3.0 similarly doubled bandwidth back when it first appeared on mainstream motherboards over 7 years ago. But how much benefit is it actually providing today? The vast majority of current graphics cards still see little to no performance benefit from running on a 3.0 x16 slot over 2.0 x16. And even if you were running one of those top-end cards that could potentially show some benefit, the rest of that 7-year old system would then be holding performance back.

And as I pointed out before, as far as real-world storage performance is concerned, most won't even notice any significant benefit from PCIe 3.0's additional bandwidth over 2.0. Plus, even if one bought a PCIe 3.0 motherboard to "future proof" themself back then, they would still need an adapter card to run modern NVMe drives, and booting from the drives might not be guaranteed. Likewise, there's no telling whether the storage devices of 5+ years from now will be fully compatible with current motherboards.

Perhaps PCIe 4.0 will be different, and there might be some tangible benefits down the line from getting a motherboard that supports it now, but I'm not sure the case is really there based on past hardware adoption trends. Either way, we're talking about mid-range motherboards here, not $200+ models, so its not like there is as sizable an investment to consider.
 
Did anyone expect to have a use for SATA3 in the future back when SATA3 was new? Not really.

A few years from now, cheap M.2-4.0 SSDs will be faster than the best available today and an upgrade to M.2-4.0 storage will be well worth it to those who have a board capable of supporting it and can be bothered with the upgrade. I do a clean Windows re-install on my PC every 3-4 years and whenever I do so, it is on a new drive so I can keep my previous install intact until I'm done re-installing everything and copying my old data over. In my case, this would likely translate to building my next system with a SATA3 SSD initially and upgrading to M.2-4.0 when I re-install Windows on it 3-4 years down the line with the expectation that NVMe M.2 will be cheaper than SATA by then due to volume, M.2 using a card edge connector (free) and not having a housing.
I remember waiting for years for USB to became useful and look at it now. In first few years nobody knew what to do with it.
 
Did anyone expect to have a use for SATA3 in the future back when SATA3 was new? Not really.

A few years from now, cheap M.2-4.0 SSDs will be faster than the best available today and an upgrade to M.2-4.0 storage will be well worth it to those who have a board capable of supporting it and can be bothered with the upgrade. I do a clean Windows re-install on my PC every 3-4 years and whenever I do so, it is on a new drive so I can keep my previous install intact until I'm done re-installing everything and copying my old data over. In my case, this would likely translate to building my next system with a SATA3 SSD initially and upgrading to M.2-4.0 when I re-install Windows on it 3-4 years down the line with the expectation that NVMe M.2 will be cheaper than SATA by then due to volume, M.2 using a card edge connector (free) and not having a housing.
They did use it, there's no dispute there as all one has to do is plug in the medium and drive to make use of it . I'm talking about full saturation, specifically the PCI-E bandwidth. The average user doesn't fully saturate SATA III on a continuous basis, let alone PCI-E 3.0. For sata III it does happen, but not enough to create an imposition towards the average end user. Lets not forget, the average end user at best deals with web browsing and word processing and even then, gaming doesn't fully saturate pci-e 3.0, yet. In time maybe, maybe not. PCI-E 4.0 is twice the bandwidth of 3.0. FYI 5.0 is in the near future which will double 4.0's bandwidth.
 
Five years from now, Zen 5 will be less than two years old, possibly still in volume production....
Zen1 is is less than two years old... still in volume? Are you hinting Zen5 may be the end of AMD's journey? because Intel is coming and they will be bringing changes that will either be just as fascinating as Zen has been or I'm going to be utterly bored.

Either way...in 5 years my system will be 5 years old and my experience with the march of tech strongly suggests a 5 year old system is rarely worth putting current day, state-of-the-art tech in and expecting to get much from it.
 

InvalidError

Titan
Moderator
They did use it, there's no dispute there as all one has to do is plug in the medium and drive to make use of it . I'm talking about full saturation, specifically the PCI-E bandwidth. The average user doesn't fully saturate SATA III on a continuous basis, let alone PCI-E 3.0.
We'll get there when more software and larger chunks of OSes get re-written to better leverage IO concurrency on fast storage. Right now, most loading is done sequentially on a file-by-file basis, leaving most cores doing nothing.

As for PCIe 5 being in the "near future", that is still going to be at least two years from now, nowhere near soon enough to have any bearing on anyone contemplating an upgrade within the next year or so. PCIe 4 on the other hand is available right now at least on AMD's side and from the entry-level if it does not choose to sabotage the CPU based on what chipset gets paired with it.