Get the Ryzen 9 3900x or wait for the 3950x

[19power97]

The 3000 series for the Ryzen line up has been a success for AMD and I'm excited to join team red by jumping all in on the latest release.

 

However, I'm conflicted in choosing the 12 core 3900x or waiting for the 16 core 3950x which is just around the corner.

 

For context, I have used the x99 platform from intel since 2015. Originally I was running the 6 core 5820k until a new corsair power supply shorted and toasted the motherboard and the CPU. A damage claim with Corsair later and I've been running the 6850k with an updated x99 board for the past 2 years until 2 months ago. Turns out after much headache and time, the 6850k died without warning on idle leaving me without my system.

 

Besides the blemishes I've had for the last 4 years, I was happily using my setup for Gaming, Editing and 3D rendering in Cinema 4D. I'd say 65%of my time leans more towards gaming and the rest goes between Premiere, After Effects, Handbrake, and Lately Cinema 4D with ProRender and Octane. Cinema 4D has been a growing interest for me for modeling and product visualizations and thus I made the investment years ago in dual 1080ti s for the purpose of expanding into the fields. I made the choice in the x99 platform for the 40 PCI Lanes to optimize the SLI setup. Unfortunately, that was also the reason I stayed on the platform for the past 4 years

 

Luckily the timing of the lastest zen 2 processors came out with the same PCI lane support and now I'm between the 3900x or the 3950x with the x570 ASUS ROG Hero VIII.

 

Do I bite the bullet and snag a 3900x somehow or hold my horse and snag the 3950x at launch considering my use case? Money isn't a consideration here.

[Billy Pilgrim]

7 minutes ago, 19power97 said:

Money isn't a consideration here.

Then I think that as long as there are no disadvantages to the 3950x over the 3900x which I believe there aren't you should go for that.

[19power97]

12 minutes ago, Billy Pilgrim said:

Then I think that as long as there are no disadvantages to the 3950x over the 3900x which I believe there aren't you should go for that.

Thing only thing that concerns me that would be a disadvantage is the reduction in base clock from 3.8 to 3.5Ghz. I understand that typically as core count goes up, base clock per single core goes down but the I'm concerned that the overclocking potential of the 3950x will be limited compared to its 12 core cousin. I've seen that most overclocks on the 3900x are generally between 4.2 to 4.4Ghz which (theoretically) could mean that the 3950x won't overclock much further than 4Ghz. The only reason I have that thought lingering in my head is that I want some kind or single-core improvement over the 4Ghz speeds my 6850k has suffered in some games.

 

I know that the assumption is about an unreleased chip but I can't afford to wait for stocks to replenish for the 3950x post-launch if I wait to see the Reviews. Don't want to end up hinging on the notification from Newegg you know?

[Billy Pilgrim]

2 minutes ago, 19power97 said:

Thing only thing that concerns me that would be a disadvantage is the reduction in base clock from 3.8 to 3.5Ghz. I understand that typically as core count goes up, base clock per single core goes down but the I'm concerned that the overclocking potential of the 3950x will be limited compared to its 12 core cousin. I've seen that most overclocks on the 3900x are generally between 4.2 to 4.4Ghz which (theoretically) could mean that the 3950x won't overclock much further than 4Ghz. The only reason I have that thought lingering in my head is that I want some kind or single-core improvement over the 4Ghz speeds my 6850k has suffered in some games.

 

I know that the assumption is about an unreleased chip but I can't afford to wait for stocks to replenish for the 3950x post-launch if I wait to see the Reviews. Don't want to end up hinging on the notification from Newegg you know?

This does seem like a hard choice. However you may be able to achieve the same clock speeds on the 3950x by disabling cores and, if money isn't and issue, investing in a beefy cooling solution. No guarantees though.

[Jimkirk363]

I was considering the 3950x but thought to myself why spend all that money when my 2700x is doing everything and more i want.

[19power97]

25 minutes ago, Billy Pilgrim said:

This does seem like a hard choice. However you may be able to achieve the same clock speeds on the 3950x by disabling cores and, if money isn't and issue, investing in a beefy cooling solution. No guarantees though.

That's a fair point. Still trying to figure out when the exact release date of the 3950x is going to be. Only thing I can find is September but if that's late September then I might be forced into finding a 3900x. Been out of the rig since late April and been working on my laptop ever since. Some projects are starting to exceed what my laptop is capable of.

[St.Nick]

Honestly the trend with every single Zen 2 processor so far is that you are going to get a 4.1 to 4.3 base clock on manual overclocking regardless of whether your using a 3600 or a 3900x and it's just up to the silicon lottery for where exactly your chip will land in that range. I expect the 3950x is going to follow the thread of that seemingly being the limit imposed by the chips architecture but more importantly pure raw clock speed just isn't the real deciding factor on how well the zen2 processors preform. I don't think the 3950x or 3900x will see a meaningful difference in your use case just because of limited  optimization for more cores. You might actually be better of with a 3900x and save the price diffrence for the lowest latency ram you can get to boost your preformance.

[lansing]

Your only choice is wait for 3950x, you're not going to get 3900x anyway because it is out of stock and is likely going to be out of stock whenever they restock again

[Triboal]

I've heard that the 3950X is going to be a binned version of the 3900X along with the two additional cores (and four extra threads obviously). Apparently this is why the launch of the 3950X was delayed but I may be wrong. In terms of overclocking, like @St.Nick saidn you should be looking to be able to hit the same clock speeds as the other Zen 2 chips when manually overclocking. 

 

I think that if you can wait from late April until now, I'm confident you can stretch out a few more weeks to make a better and more informed decision.

[BigDamn]

The lower base clock of the 3950x will hurt its performance in Adobe software, which rely on higher clock speeds. How significant of a hit the 3950x will take is yet to be seen but I can't imagine it'll be too damning. As for rendering software like Cinema4D the 3950x will be noticeably faster than the 3900x.

 

If overclocking is an option for you I'd go with the 3950x. If you're able to push it upwards of 4.0-4.1ghz you won't see a noticeable dip in single-threaded performance and you'll have a great multi-threaded advantage. If overclocking isn't an option than the 3900x may be the better overall chip for you. Wait until benchmarks release and make your decision based on that.

[HeyYouOK?]

3 hours ago, lansing said:

Your only choice is wait for 3950x, you're not going to get 3900x anyway because it is out of stock and is likely going to be out of stock whenever they restock again

My local Microcenter literally has 10+ in stock right now and has had it in stock for months...I also saw Newegg stock go back up last week but they were selling for like 600+ and i just lol'ed.

[vanished]

22 hours ago, 19power97 said:

Thing only thing that concerns me that would be a disadvantage is the reduction in base clock from 3.8 to 3.5Ghz. [...]

Personally I'm expecting the boost clock to follow the trend of the existing products - ie, be even higher than the 3900x, and thus overclock better as well.  The base clock may be lower for TDP reasons but I fully expect that you'll be able to achieve an all-core boost of the same or better, with the same holding true for a single core.  If money is no object then it would be a better option because obviously it's more powerful.

22 hours ago, 19power97 said:

[...]Turns out after much headache and time, the 6850k died without warning on idle leaving me without my system.[...]

I'd like more details about this if you would be so kind.  I've literally never seen or heard of a CPU dying from "natural causes" before in over 10 years of being into this stuff on forums, either personally, or from any one of the thousands of other posts I've come across.  Was it natural causes, or was there another power spike, or was it heavily overclocked, etc.?

[19power97]

1 hour ago, Ryan_Vickers said:

 

 

 

 

1 minute ago, Ryan_Vickers said:

I'd like more details about this if you would be so kind.  I've literally never seen or heard of a CPU dying from "natural causes" before in over 10 years of being into this stuff on forums, either personally, or from any one of the thousands of other posts I've come across.  Was it natural causes, or was there another power spike, or was it heavily overclocked, etc.?

@Ryan_VickersSure. This will be long so sit back and enjoy.

 

 I had received the 6850k from Newegg after being reimbursed for the first Motherboard (the ASUS X99 Deluxe) and CPU (the 5820K) from the damage claim from Corsair after a Power Coil Shorted while under load. I had also purchased the refreshed version of the motherboard (the X99 Deluxe v2) out of the same claim. This was all in July of 2017. I attempted a modest overclock by increasing the multiplier to 38~41 and a slight increase to the stock voltage never exceeded 4Ghz. After 2 days, I discovered the chip wouldn't exceed 4Ghz, no matter the overclock on all cores or individually (the goal was 4.1 to 4.2 based on checks online for other results). I even attempted disabling the turbo boost but nothing ever resulted in a better clock speed. With all the trouble, I decided to reset to factory settings and run the chip at stock since the turbo was doing better than any overclock that was attempted.

 

Fast forward to April, the chip has still been running at stock, especially after updating the BIOS for the Spectre and Meltdown patches. Again most of the chips uptime was during gaming sessions, editing sessions, and the occasional CPU render in C4D. Turned the PC on April 28th 2019 for a several updates on the machine all ranging from standard windows 10 updates to Adobe CC updates. I was on my laptop in the same room while I let the updates insue and after 3 hours the System lost all power as if the power button held been held for 10 secs. I got up, walked to the PC and attempted to power on the system from the case power button. Once pressed, the system would go into a power cycle every 8~9secs without POSTing. I then unplugged from the wall and pressed the power button again to run the power completely out and the system still refused to power on with a Successful POST. I switched power outlets and even different power circuits in the house with no luck.

 

Then I let the system sit for several hours and attempted to power on the system once more. The system would then perform a success POST but the system would still abruptly shutdown after 20~40secs with the average being around 26sec from the record I was keeping. The System could go into the BIOS or even begin loading windows but would shutdown no matter the screen or process the system was currently in. I also recorded the Q-Codes the board would display as it cycled through its boot sequence but there was never a code the reliably display at the moment of shutdown. I can not accurately provide a guess of which code I would pick if any because of the almost random code the system would end with. I began removing RAM sticks, reinserting different sets, Clearing CMOS (both via button and battery removal), removing graphics cards and the PCI slots the master card was in. No Luck. I unplugged the AIO water cooler to see if a small short was on the pump header for 2 of the runs but again no luck. I purchased a new power supply and fully replaced all cables with no storage drives (both M.2 and Sata) with the same issue reoccurring despite my efforts to diagnose the issue. This continued for the next 3 weeks when I had spare time to diddle with the system. There were times where I would leave the system along from 5~6 hours where the system would boot and load windows for up to hour before losing power once again. During that time the system would be completely usable.

 

By May 27th 2019, I had replaced everything with exception to the motherboard and the CPU. After zero luck attempting to diagnose the problem, I could only conclude the CPU or the motherboard was at fault. I then decided to call bothe ASUS and Intel support to begin the RMA process for both components. Each had a 3-year limited warranty of which I still qualified for.

 

I shipped out the motherboard to ASUS on the 30th of May after bantering with support.

ASUS determined there to be a bent pin and sent the motherboard back. I, on the other hand, took a photo of the socket incase ASUS refused to reair the board due to shipping. I hadn't been called prior to ASUS returning the board with their results unlike I had been told before shipping the board.

I humored the bent pin diagnosis from ASUS and rebuilt the system.

 

I powered on the board once again and the same issued occurred as prior to shipping the board. No change in behavior and I began repeating all the same steps as I had done before.

 

2 hours after going through my troubleshooting as before, I cleared the CMOS again and after doing that the board finally displayed a stable Q-Code of 00. 00 being CPU Error/ No CPU found.

 

At this point I'm frustrated as I had been messing with the system for over a month and have nothing to show for it. I called ASUS, let them know what happened, and after a heated exchange, ASUS wanted the board sent back to them for further analysis.

 

I also emailed Intel for an RMA to begin shipping the now presumed dead 6850K. Intel decided to issue a new 6850k from their parts room 2 days after receiving my original 6850K.

 

ASUS on the other hand, continued to stand by their board after 14 calls over the next week and a half stating they cannot replicate my issues. They eventually sent back the same board stating no faults.

 

I requested that statement in writing along with the tests performed. By this point, I want the motherboard out of my hair and prepared to sell it on Ebay with ASUS's signature and warranty that was still valid. It sold 2 weeks ago for $180 and the buyer reported back to be fully operational.

 

The new 6850K from the RMA I specifically didn't want to open because I noticed they were still reselling for more than what it was originally priced for 2 years ago. I also listed it on Ebay and it sold 3 weeks ago for $480 which was fantastic since it was bought during that original damage claim for $359.99.

 

And that brings us to today. A naked system of 1080TIs and case waiting for a new x570 motherboard and either a 3900x or 3950x. Goes without saying that this has been quite the rollercoaster of issues and bantering but I'm ready to put this all aside and get things running under team red. I'm ready to have my system back after all these months.

 

 

 

[vanished]

18 minutes ago, 19power97 said:

@Ryan_VickersSure. This will be long so sit back and enjoy.

 

[snip]

Wow that is quite a story, and amazingly it does seem like the CPU actually did die, and while at stock no less.  Well, congrats I guess, that is the first time I've ever seen this happen I'm glad you did well with it in the end despite that and weren't out of pocket.

[19power97]

10 minutes ago, Ryan_Vickers said:

Wow that is quite a story, and amazingly it does seem like the CPU actually did die, and while at stock no less.  Well, congrats I guess, that is the first time I've ever seen this happen I'm glad you did well with it in the end despite that and weren't out of pocket.

Glad I could be the first I suppose. I never thought it could be the CPU for the majority of the RMA process as there were no signs of wear or issues prior to the system failing. I originally thought the power supply failed again which was the Corsair RM1000x. However, once I replaced it with an off the shelf new inbox of the same model and had the exact same issue, I was in confusion as most of the patterns I was seeing was random. I could only assume that the issue was heat-related. Either in the form of a short in the CPU itself or with the power delivery of the motherboard but I used a heat gun to measure points of interest on the board and nothing went above 34C. The CPU measured at 38C at most in the BIOS without any peaks prior to shutdown but I'm left with no other explanation prior to it completely failing.

[bignaz]

I wouldn't even look at a 3950x unless you are on a custom highend loop. Even the 3900x man handles aio's and can give a highend custom loop a good workout. These cpu's run hot.

[19power97]

18 minutes ago, bignaz said:

I wouldn't even look at a 3950x unless you are on a custom highend loop. Even the 3900x man handles aio's and can give a highend custom loop a good workout. These cpu's run hot.

I'm willing to consider reconfiguring my cooling setup to accommodate the increase in temperature. Right now I'm running all Noctua Fans in Push-Pull Configuration with 9 fans total cycling air. A corsair H115i is installed currently as well but I've been looking to replace that soon likely with the change in CPU. Appreciate the heads up.

[vanished]

49 minutes ago, bignaz said:

I wouldn't even look at a 3950x unless you are on a custom highend loop. Even the 3900x man handles aio's and can give a highend custom loop a good workout. These cpu's run hot.

It's heavy but it's less than a 9900k and plenty of people seem to manage them just fine

[St.Nick]

Honestly I'm with Ryan on this one. You could just drop a NH-D15 in there if you're already that invested in Noctua and be fine on air. If it can tame an overclocked FX-9590  the 3950X will be fine too. 

[lansing]

5 hours ago, HeyYouOK? said:

My local Microcenter literally has 10+ in stock right now and has had it in stock for months...I also saw Newegg stock go back up last week but they were selling for like 600+ and i just lol'ed.

Those stocks on Newegg are from 3rd party sellers, not Newegg, that's why you see the jacked up price. The Newegg stock got snipe in 30 minutes every time it went back up.

[xg32]

2 hours ago, 19power97 said:

.

IMHO i have no interest in the 3900x for a workstation knowing that a 3950x is on the way. However, i have mild expectations for what the clocks are gonna be, the whispers were 300w on 4.3 all core boost, so with a 150w-180w cooler like a d15 or DRP4, i'm expecting 4.0-4.1 for daily workloads. Which is fine for a 16 core part. I do plan to spend a bit more on binned b-die ram though, it's worth it.

[vanished]

46 minutes ago, xg32 said:

IMHO i have no interest in the 3900x for a workstation knowing that a 3950x is on the way. However, i have mild expectations for what the clocks are gonna be, the whispers were 300w on 4.3 all core boost, so with a 150w-180w cooler like a d15 or DRP4, i'm expecting 4.0-4.1 for daily workloads. Which is fine for a 16 core part. I do plan to spend a bit more on binned b-die ram though, it's worth it.

A 3900X at stock is apparently around 142 W (source below).  Another 33% to get it to 16 cores would bring that to 189 W, assuming the silicon is of equal quality, and if the existing lineup is anything to go by, it won't be equal, it will be better, so I'd like to see where the 300 W rumours are coming from and what that's based on because I've not heard that before and it doesn't make sense to me with the information I currently have.

 

Spoiler

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/19

[xg32]

1 minute ago, Ryan_Vickers said:

A 3900X at stock is apparently around 142 W (source below).  Another 33% to get it to 16 cores would bring that to 189 W, assuming the silicon is of equal quality, and if the existing lineup is anything to go by, it won't be equal, it will be better, so I'd like to see where the 300 W rumours are coming from and what that's based on because I've not heard that before and it doesn't make sense to me with the information I currently have.

 

  Reveal hidden contents

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/19

this was during computex when tech jesus said the chip wasn't ready, and i doubt the 3900x all core 4.3 is actually 142w, it's likely higher since thats the wall, gonna have to do some research on that myself

[vanished]

9 minutes ago, xg32 said:

this was during computex when tech jesus said the chip wasn't ready.

Well in that case then maybe his review is more reliable

 

Looks the same to me though

 

9 minutes ago, xg32 said:

this was during computex when tech jesus said the chip wasn't ready, and i doubt the 3900x all core 4.3 is actually 142w, it's likely higher since thats the wall, gonna have to do some research on that myself

Well generally they do 4.2 - 4.3 all core at stock from what I've seen, but manually pushed here he got it up to 170 W, which is still only 227W scaled up to 16 cores in theory.  Also, in this case wall power doesn't matter, it's actually what's coming right off the chip, since we're only measuring power for the sake of getting a sense of heat.  But if if this was wall power then the chip would actually be less not more since wall power is always greater due to PSU inefficiency.

[xg32]

4 minutes ago, Ryan_Vickers said:

.

if the 3950x does 4.1-4.2 at 175w then that's already beyond my expectations, ill take it lol