AMD Ryzen R9 3900X Review Round Up

Zarathustra[H] said:
Well, you do realize that the only reason we got Bulldozer is because AMD didn't have enough money to spend on R&D, and the reason for that was that Intel illegally blocked them out of OEM sales and sabotaged performance in their popular compiler.

Sure, they settled for a billion dollars, but that was a fraction of the damage Intel did. They just had a near bankrupt AMD over a barrel and more or less forced them to settle for peanuts because they needed the money as soon as possible.

In the absence of competition - however - Intel have given us generation after generation of embarrassingly bad single digit performance increases.

I welcome this turn of events, and although AMD may not be quite as fast as Intel for gaming purposes, I look forward to buying one of their chips.

The true evil here is Intel. They have been a shady company from day one, with no qualms about using shameful litigation tactics to try to kill any competition.

Intel and the x86 Architecture: A Legal Perspective

I have reluctantly been using Intel for several years now, as I hated giving a shitty company like that my money, and finally I don't have to anymore.
Click to expand...

Forget Fermi, FX 5800 was true original dustbuster, and the entire FX series as a whole was kinda lackluster. To this day I still can't figure out how anyone at nVidia OK'd that dental drill of a fan on the FX 5800. I mean yeah sure we've had loud blowers from both camps after that, but nothing even came close to the ear-piercing high pitched shrill of the FX 5800 fan.

Derangel said:
ryan_975 said:
As far as milking the mainstream with low core counts and new sockets/chipset goes.. Intel definitely deserves all the criticism they get.

However, regarding single digit performance gaines, since AMD hasn't actually surpassed Intel yet, I'm not yet convinced that Intel hasn't wrung x86 for almost everything they can get. We need to wait and see if AMD can actually surpass Intel and kick off some real performance leapfrogging to know if they were holding back.
Click to expand...
Zen 2 IPC is pretty close to, if not above, current gen Intel stuff across the board. The big reason Zen 2 doesn't win out across the board is simply clock speed. Intel is promising big IPC gains from their Icelake mobile chips. If that holds true on the 10nm desktop parts I think we'll have proof of Intel resting on its laurels.
Click to expand...

Didn't some leaked Geekbench scores purportedly show Sunny Cove having +18% IPC over Skylake? If true then you know Intel has been majorly sandbagging big time.
 
Derangel said:
Zen 2 IPC is pretty close to, if not above, current gen Intel stuff across the board. The big reason Zen 2 doesn't win out across the board is simply clock speed. Intel is promising big IPC gains from their Icelake mobile chips. If that holds true on the 10nm desktop parts I think we'll have proof of Intel resting on its laurels.
Click to expand...

It's not just clocks. Getting to 5ghz still wouldn't bring it to complete parity, but it would be close enough and of course dominate in multi-threaded.

I'm hoping AMD rewards consumers with their loyalty by giving us a Zen2+ with higher clocks that is mostly compatible with current chipsets.
 
Derangel said:
All companies make bad products. I see Netburst didn't turn you away from Intel nor did Fermi turn you away from Nvidia.
Click to expand...

Ha, this guy holds the record for the stupidest post in this thread, that's for sure. Bulldozer has nothing on the piece of shit Pentium D 820 and its ilk.
 
Last edited:
dgz said:
Ha, this guy holds the record for the stupidest post in this thread, that's for sure. Bulldozer has nothing on the piece of shit Pentium D 820 and its ilk.
Click to expand...

Netburst was an Epic fail, especially when tied to Rambus RAM (though not quite as bad as Bulldozer which had less then half the per core performance of the competition at the time in many workloads, and actually moved backwards in IPC compared to Phenom II), but what did you have against Fermi?

Both the GTX 480 and the GTX 580 were the fastest GPU's money could buy when they launched, and they were great products.

I had a GTX 470 that overclocked like a bat out of hell back then, and upgraded to a 580 after the refresh. No regrets.

Sure, they ran a little hot, but they were pushing the limits of what the 40nm process at the time could do.
 
Zarathustra[H] said:
Netburst was an Epic fail, especially when tied to Rambus RAM (though not quite as bad as Bulldozer which had less then half the per core performance of the competition at the time in many workloads, and actually moved backwards in IPC compared to Phenom II), but what did you have against Fermi?

Both the GTX 480 and the GTX 580 were the fastest GPU's money could buy when they launched, and they were great products.

I had a GTX 470 that overclocked like a bat out of hell back then, and upgraded to a 580 after the refresh. No regrets.

Sure, they ran a little hot, but they were pushing the limits of what the 40nm process at the time could do.
Click to expand...

The 580 was great, the 480 was hot garbage. It was technically "the fastest" but not exactly by a huge margin over the 5870. At the time, the 5870 was simply the better card to buy. The 470 was better but, again, the AMD competition was the better purchase at the time.

They didn't run "a little hot". They ran extremely hot, to the point where it required water cooling to keep temps under control. The 580, on the other hand, fixed pretty much everything wrong with the 480. To quote Brent's review: The 580 is what the 480 should have been. Nvidia rushed the 580 out in just under seven months from the release of the 480. Even they wanted people to forget about first gen Fermi.
 
Zarathustra[H] said:
Well, you do realize that the only reason we got Bulldozer is because AMD didn't have enough money to spend on R&D, and the reason for that was that Intel illegally blocked them out of OEM sales and sabotaged performance in their popular compiler.

Sure, they settled for a billion dollars, but that was a fraction of the damage Intel did. They just had a near bankrupt AMD over a barrel and more or less forced them to settle for peanuts because they needed the money as soon as possible.

In the absence of competition - however - Intel have given us generation after generation of embarrassingly bad single digit performance increases.

I welcome this turn of events, and although AMD may not be quite as fast as Intel for gaming purposes, I look forward to buying one of their chips.

The true evil here is Intel. They have been a shady company from day one, with no qualms about using shameful litigation tactics to try to kill any competition.

Intel and the x86 Architecture: A Legal Perspective

I have reluctantly been using Intel for several years now, as I hated giving a shitty company like that my money, and finally I don't have to anymore.
Click to expand...

Agreed

With AMD being so close, People have no reason to go to Intel. We should stop giving intel our money.
 
Last edited:
I have recently gone to a 3600X, i went with the X as the promo on it was only just more than the 3600 and i don't overclock and the Wraith Spire RGB cooler with the X is aesthetically pleasing with the Gigabyte Aorus B450 I board I got second hand for rather cheap. I was running a 4790K/980ti setup and I have noticed much better performance, less stuttering in Rust and Battlefield 5, notably in Rust the 3600X is so much less prone to hitching even running stock. My biggest surprise was how well the GTX 1660ti performs relative to my old 980ti, the maxwell aged badly as the 1660ti matches or flat out beats the 980ti particularly in Rust. I am happy with the upgrade.
 
  • Like
Reactions: N4CR
like this
Mchart said:
It's not just clocks. Getting to 5ghz still wouldn't bring it to complete parity, but it would be close enough and of course dominate in multi-threaded.

I'm hoping AMD rewards consumers with their loyalty by giving us a Zen2+ with higher clocks that is mostly compatible with current chipsets.
Click to expand...

Stilts IPC comparison shows that it's all down to clocks (and avx 512).

For the first time in over a decade, AMD has reached IPC parity with Intel.
On average, based on the results of 32 individual workloads Zen 2 even manages to provide slightly higher average IPC than Coffee Lake-S Refresh.
Thanks to it AVX-512 resources Skylake-X manages to stay a head in this test suite however, not by a large margin.
Click to expand...

https://www.overclock.net/forum/10-...nical-matisse-not-really.html#/topics/1728758
 
Any idea when to expect these CPUs to actually be in stock? I'm too busy to constantly check newegg and Amazon. Signed up for email notifications on 3700x-3800x.

I only upgrade every half decade.. haven't seen things sell out this way in ages, either yields are low or desktop pc enthusiasm at an all time high.
 
Nolan7689 said:
Well, the way I’m reading those charts is that post update the 7600 is not faster. Am I misreading something?
Click to expand...

The 7600k sits right up with the top-end Ryzen CPUs at 1080p, and above all of them at 1440p.

Nightfire said:
No you are not misreading. Our friend here is just trying to marginalize the improvements in one of the worst performing games for Ryzen.
Click to expand...

Do not assume my intentions.

My point is that the shared benchmarks more show that the game itself is not very CPU dependent, and to the extent that it is, it is clockspeed dependent more than multithread dependent.

So it seems like a poor example for a CPU performance comparison, since it will run well on anything.
 
n=1 said:
If true then you know Intel has been majorly sandbagging big time.
Click to expand...

Intel expected to have Ice Lake (and the Sunny Cove mobile SoC) out three or four years ago when they had originally planned for 10nm mass production. Given that they're losing marketshare to AMD not just because AMD presents a decent performance argument but also because Intel literally can't ship enough processors to meet demand, it's pretty clear that Intel has been sandbagging themselves too.
 
Mode13 said:
Any idea when to expect these CPUs to actually be in stock? I'm too busy to constantly check newegg and Amazon. Signed up for email notifications on 3700x-3800x.

I only upgrade every half decade.. haven't seen things sell out this way in ages, either yields are low or desktop pc enthusiasm at an all time high.
Click to expand...
I just looked on Amazon, 3700x appears in stock for me.

IdiotInCharge said:
The 7600k sits right up with the top-end Ryzen CPUs at 1080p, and above all of them at 1440p.
Click to expand...

In the pre-patch charts yes it does, post patch the 7600k wasn’t shown. And given that it appears the Intel parts shown didn’t change much with the patch it stands to reason that the results of the 7600k also didn’t change much and is now lower than the Ryzen parts shown.
 
Nolan7689 said:
In the pre-patch charts yes it does, post patch the 7600k wasn’t shown. And given that it appears the Intel parts shown didn’t change much with the patch it stands to reason that the results of the 7600k also didn’t change much and is now lower than the Ryzen parts shown.
Click to expand...

Well, all of the AMD CPUs were below all of the Intel CPUs on pre- and post-patch charts at 1440p.
 
IdiotInCharge said:
Well, all of the AMD CPUs were below all of the Intel CPUs on pre- and post-patch charts at 1440p.
Click to expand...

This all leaves the question: How many other games favor intel in such a manor and don't get patched to properly support AMD? I am not saying that is the case, but we have no real way of knowing.
 
  • Like
Reactions: N4CR
like this
NWRMidnight said:
This all leaves the question: How many other games favor intel in such a manor and don't get patched to properly support AMD? I am not saying that is the case, but we have no real way of knowing.
Click to expand...

This also assumes that patching the games can shift performance the other way. Some games may simply never benefit from additional optimization that favors AMD.
 
Mode13 said:
Any idea when to expect these CPUs to actually be in stock? I'm too busy to constantly check newegg and Amazon. Signed up for email notifications on 3700x-3800x.

I only upgrade every half decade.. haven't seen things sell out this way in ages, either yields are low or desktop pc enthusiasm at an all time high.
Click to expand...

I imagine they'll be in and out of stock for a couple to a few weeks before you're able to find them reliably. I saw the 3700x pop up for a little bit on Egg a couple days ago, but I don't think it lasted long.
 
Zarathustra[H] said:
Does anyone know if the current AMD AGESA shipping with most x570 motherboards will post with a 3950x when they ship?

I'm tempted to order my motherboard now, so I make sure I get the one I want, but I don't want to wind up being stuck with a board I need to flash and no CPU to flash it with come September.
Click to expand...

Derangel said:
The CPU support list for boards should mention whether they support the 3950X or not. If not there are quite a few X570 boards with BIOS Flashback so you won’t need a CPU installed to update. I think MSI and Gigabyte have it along most of their X570 line.
Click to expand...


To answer my own question, I asked this in a support email to Asus.

The relevant part of their response was as follows:

ASUS Support said:
As per our product engineers, any X570 chipset board will support the 3rd generation Ryzen CPUs out of the box. It will not require any BIOS updates for this to work with the 3950x.
Click to expand...
 
I would not buy a MB now for a build happening months in the future.
Why?
Because a lot of MB vendors revise their boards (silkscreen v1.01, etc.) and the improvements could be nice to have or might even be somewhat important to have. For example, the very early C7H boards had an issue with voltage readings in software being off by 0.05V that was corrected in a revised version of the board. Details about this issue are in this post: https://rog.asus.com/forum/showthread.php?101617-Crosshair-VII-Hero-Essential-Info-Thread
 
Azrak said:
I would not buy a MB now for a build happening months in the future.
Why?
Because a lot of MB vendors revise their boards (silkscreen v1.01, etc.) and the improvements could be nice to have or might even be somewhat important to have. For example, the very early C7H boards had an issue with voltage readings in software being off by 0.05V that was corrected in a revised version of the board. Details about this issue are in this post: https://rog.asus.com/forum/showthread.php?101617-Crosshair-VII-Hero-Essential-Info-Thread
Click to expand...

This is true, but revisions like that aren't all that common. Most of the time, whatever v1.0 PCB you get is what you will see throughout the life time of that product. There are periods where this is more common, such as back during the 680i SLI days where even the Intel boards were revised a lot. You saw this more with GIGABYTE than anyone. These days, I haven't seen it all that much. The one that really needed it was the Crosshair VI series. We did have three of them die on the test bench during the HardOCP reviews of them.
 
Azrak said:
I would not buy a MB now for a build happening months in the future.
Why?
Because a lot of MB vendors revise their boards (silkscreen v1.01, etc.) and the improvements could be nice to have or might even be somewhat important to have. For example, the very early C7H boards had an issue with voltage readings in software being off by 0.05V that was corrected in a revised version of the board. Details about this issue are in this post: https://rog.asus.com/forum/showthread.php?101617-Crosshair-VII-Hero-Essential-Info-Thread
Click to expand...

That is something I have thought of, and I am still debating it. I'm just concerned that once launch quantities are depleted, there may be shortages.

I'm also concerned RAM and SSD's are going to spike back up in price with everyone building new Ryzen systems. They had finally come down to a somewhat tolerable price level.

Ideally I'd like to have every other components available so that I can order a 3950x on lunch day and start building as soon as it arrives. I've been putting this upgrade off for years. I'm really itching to get it done.
 
Dan_D said:
This is true, but revisions like that aren't all that common.
Click to expand...
Not trying to argue with you but MSI just updated much of their MB lineup to switch to 32 MB BIOS sizes. https://www.techpowerup.com/257240/...400-series-motherboards-with-256mb-bios-chips
I'll have to take your word that revisions are rare since I only buy one board during its production lifetime. I just know that in the past, I've seen many versions of MBs that I own mentioned in forums and I usually have the earliest version of the board much to my chagrin.

Zarathustra[H] said:
I'm just concerned that once launch quantities are depleted, there may be shortages.
Click to expand...
That is understandable.

Zarathustra[H] said:
I'm also concerned RAM and SSD's are going to spike back up in price
Click to expand...
This is a real possibility. Also, if you were thinking about getting B die memory, better to get it now before stock runs out AND prices are low-ish.
 
  • Like
Reactions: N4CR
like this
It remains to be seen, but early boards should be fine from MSI. Right now the only feature lost due to BIOS size is RAID and they claim to be adding it back ASAP. They also have stated that they plan on re-releasing the full GUI BIOS that drops support for everything pre-3000 series. Essentially, the current stripped down BIOS is just a bridge for pre-3000 series users. Once you update you switch CPUs, update again, and you haven’t lost anything.

I wouldn’t worry about buying a MB now is what I’m trying to say.
 
Azrak said:
Not trying to argue with you but MSI just updated much of their MB lineup to switch to 32 MB BIOS sizes. https://www.techpowerup.com/257240/...400-series-motherboards-with-256mb-bios-chips
I'll have to take your word that revisions are rare since I only buy one board during its production lifetime. I just know that in the past, I've seen many versions of MBs that I own mentioned in forums and I usually have the earliest version of the board much to my chagrin.


That is understandable.


This is a real possibility. Also, if you were thinking about getting B die memory, better to get it now before stock runs out AND prices are low-ish.
Click to expand...

As I said, PCB revisions are rare on production motherboards. However, I did say that it does happen during certain periods. Specifically, in cases like the one you brought up with the BIOS chips. But it isn't routine in the sense that there are many generations of motherboards from most vendors where you only ever saw one PCB revision out in the wild. However, I've seen everything from heat sink updates to VRM changes on various revisions of a given motherboard. Creative Labs is the king of PCB revisions. Although, with them you are generally better off with earlier versions as they tend to strip out features and make them cheaper as time goes on.

I've actually owned more than one PCB revision of a motherboard. I never meant to say that it didn't happen, just that it's relatively rare given how many models exist. I've also actually had first run and last run copies of a given board. That was the case with my D5400XS and Rampage IV Formula. In some of these cases, I never saw any changes to the PCB. The D5400XS changed, but the Rampage IV Formula didn't. I've also seen slight changes to component selection even if the PCB revision markings never changed.

Mchart said:
I wouldn’t worry about buying a MB now is what I’m trying to say.
Click to expand...

I wouldn't either.
 
  • Like
Reactions: N4CR
like this
Dan_D said:
I've also seen slight changes to component selection even if the PCB revision markings never changed.
Click to expand...

From my limited experience with the board industry (boards for custom medical equipment, not PC Motherboards) usually these changes have to do with parts obsolescence. There are so many little components on a board, and not all of them have long term lifespans, and their replacements may be subtly different to the point where they are not a drop in replacement. You often have to re-spin boards, not because something was wrong with the old ones, or because you are adding features, but simply in order to make the design work with the new replacement parts work.

This was huge during the implementation of the European ROHS directive which - among other things - banned the use of lead in most products causing electronics manufacturers to scramble. It was technically passed in 2003 and went into effect in 2006, but there were so many loop holes in the original version that implementation on many products was delayed and hit different companies depending on how their legal departments interpreted the regulation in different years.

In some cases previous leaded parts were discontinued in favor of parts without lead, causing board makers to scramble. In other cases the board makers were taking the initiative to produce ROHS compliant boards on their own initiative.

I've seen several board designs first hand that just stop working once you replace the old leaded parts with newer unleaded ones, having to be troubleshot and respun.
 
  • Like
Reactions: N4CR
like this
Zarathustra[H] said:
From my limited experience with the board industry (boards for custom medical equipment, not PC Motherboards) usually these changes have to do with parts obsolescence. There are so many little components on a board, and not all of them have long term lifespans, and their replacements may be subtly different to the point where they are not a drop in replacement. You often have to re-spin boards, not because something was wrong with the old ones, or because you are adding features, but simply in order to make the design work with the new replacement parts work.

This was huge during the implementation of the European ROHS directive which - among other things - banned the use of lead in most products causing electronics manufacturers to scramble. It was technically passed in 2003 and went into effect in 2006, but there were so many loop holes in the original version that implementation on many products was delayed and hit different companies depending on how their legal departments interpreted the regulation in different years.

In some cases previous leaded parts were discontinued in favor of parts without lead, causing board makers to scramble. In other cases the board makers were taking the initiative to produce ROHS compliant boards on their own initiative.

I've seen several board designs first hand that just stop working once you replace the old leaded parts with newer unleaded ones, having to be troubleshot and respun.
Click to expand...

You are correct. Sometimes PCB revisions are made for the reasons you stated. It's much more common on products with longer life spans. Motherboards generally have a short window of sales before being EOL.
 
  • Like
Reactions: N4CR
like this
Zarathustra[H] said:
Does anyone know if the current AMD AGESA shipping with most x570 motherboards will post with a 3950x when they ship?

I'm tempted to order my motherboard now, so I make sure I get the one I want, but I don't want to wind up being stuck with a board I need to flash and no CPU to flash it with come September.
Click to expand...
I think you should wait to get your board until a new revision comes out. What if they add a bunch of RGB to the board in the next revision, you wouldn't want to miss out on that right?
 
Zarathustra[H] said:
Does anyone know if the current AMD AGESA shipping with most x570 motherboards will post with a 3950x when they ship?

I'm tempted to order my motherboard now, so I make sure I get the one I want, but I don't want to wind up being stuck with a board I need to flash and no CPU to flash it with come September.
Click to expand...

We have absolutely no idea. The best thing you could do if you want to buy the board now is to buy one that supports flashing without a CPU or RAM installed.
 
Dan_D said:
We have absolutely no idea. The best thing you could do if you want to buy the board now is to buy one that supports flashing without a CPU or RAM installed.
Click to expand...

If you look up a few posts, I contacted ASUS support. They claim their engineers say that it should be out of the box compatible with any current or future Ryzen 3000 chips, presumaböu provided there are no major architectural changes.
 
Zarathustra[H] said:
If you look up a few posts, I contacted ASUS support. They claim their engineers say that it should be out of the box compatible with any current or future Ryzen 3000 chips, presumaböu provided there are no major architectural changes.
Click to expand...

That's probably true as the Ryzen 9 3950X is already known to exist and we know when it will be available. Microcode could change significantly, but I wouldn't bet on it. That said, you will almost certainly see newer UEFI BIOS revisions that include official support for the CPU right before it launches.
 
Mode13 said:
Retail price or third party seller asking $100 more??? I can't find them for $329
Click to expand...


You should have bought them during launch week, before the supplies ran out. I saw both 3900x and 3700x in-stock at amazon at various points (I'm sure that all disappeared now that Prime day is on, and people are looking to buy things after the extended holiday many took)

Also, newegg still has MSRP stock of 3800x:

https://www.newegg.com/amd-ryzen-7-3800x/p/N82E16819113104

Also, check your local Microcenter. Mine had 10+ of the 3900X on the day after launch day (sold-out now).

Or you can just settle for the 3600x? You can buy that anywhere., and it has 95% of the single-threaded performance of the 3700x

You're always going to find shortages of high-end processors on launch week. They can only build up so much stock before it starts costing AMD serious money to sit on that inventory, and the demand tends to be largest in the first few months of release. So they intend to meet demand eventually.
 
Last edited:
You must log in or register to reply here.
Back
Top