AMD announced the pricing and availability for Threadripper

,

Any word on Threadripper motherboard pricing?

Is it expected to be comparable to x299 or is it going to be more in the server motherboard price range?

2 Likes

Probably not that cheap with all them PCIe lanes. :wink:

Probably similar or a bit higher than X299, but not by much I assume.

1 Like

X399 boards arent going to be cheap.
I kinda suppose arround $300,- and upwards.

So far the most common guesses I've heard reckon that it will be around the same price point as X299. There are more PCIe lanes but their configuration shouldn't be as difficult as X299. The CPU socket is larger, but everything else seems rather similar to X299 in my opinion.

I just hope they will come with a decent power delivery, since the chips have a slightly higher TDP than e.g. the i9-7900X and X299 is a bit of a fail when it comes to that.

Well that X299 vrm issue story is really blown out of the water a bit.
I have not seen manny other complaints about it, wenn it comes to the more decent boards.

As far as the vrm implementations on X399 are concerned.
Most boards will have decent enough design.
According to Gigabyte who had some of their X399 pre production models at computax.
Those boards were 6 phase boards with 40A IR powerstages.
But according to Gigabyte those boards were just pre production models.
Their final lineup will be 8 true phases.

Asus will also come with 8 true phase designs.

And Asrock is probablly either comming with 8 true phase design aswell.
Or maybe 6 phases doubled to 12 or something like that.
But never the less those boards should be more then fine.

1 Like

After watching der8auer's video, I think it's bad enough. You can't really overclock with those boards and I don't even want to imagine how the higher core-count CPUs will be like, since they will probably have a higher TDP too.

Since X399 is less rushed and the board manufacturers knew that Threadripper will be more power hungry, I can imagine that they had better components in mind from the start. So I think the chances are good.

Well i dont know which particular boards he has tested.
But 8 true phases for Vcore with 60A IR powerstages is pretty much the highest you could get without a doubling scheme.
And with boards like that it really shouldnt be an issue.
I asssume that boards like the Asus Rampage VI Extreme will have a similar layout like that.
And that should be more then fine really.

But yeah buying the top end cpu, and then go with the cheapest motherboard you could find.
Is never really the best thing to do.
Also the issues that Der8auer was refering to in his video,
might have been an heatsink design issue, and not a vrm issue persee.

1 Like

Point is that if this is already an issue with a middling OC (4.6GHz) on a cherry-picked 10-core, imagine what it will mean for the 14+ core CPUs. Der8aeur tested ASUS prime x299 A (e280), Aorus gaming 3 (e290), and msi pro gaming carbon (e320), none could sustain 4.6GHz OC on the 7900x. Yes, they're "entry" level boards, but still, no way they're going to work with the biggest CPUs -- yet they are marketed as capable of driving the entire line-up. (And in this specific lineup, the 7900x isn't really "top-end" -- those would be the 14-18c parts.)

1 Like

I am starting to see confirmation of something I suspected on the whole X299 overclocking thing: There just aren't enough pins to deliver enough power even if you can manage the heat that X299 generates. We're starting to see a lot of reports of scorched pins from dumping > 250 watts sustained into a 140 watt tdp cpu.

I have so far tested the MSI SLI Plus and the gigabyte X299 gaming 9. The MSI SLI plus has a single 8 pin but can deliver ~ 350w to the cpu before the VRMs pass 85 degrees C (whole setup is raiting for 125c) with practically no airflow.

The gigabyte gaming 9 has two 8 pin CPU power connectors and the VRMs run quite a bit cooler -- have a heat pipe design -- but once again, I dont think dumping that much power into the cpu is really feasable. At least for the 10 cores that I've been testing. 250 watts is about the safe max sustained with peaking at maybe just maybe 300w for like 15 seconds? but thats realllly pushing it.

Maybe Asus will change my mind when testing.. we'll see. I've got the Strix X299 and its one four pin plus one 8 pin. Raw wire power that's like 500 watts before the connectors really start heating.. I can't imagine dumping 500 watts into a 140w tdp cpu and expecting it to do much. It can be stable if delidded, but I'd love to know the actual specs on this socket. I bet it is around 205 watts, but I still have some homework to do.

Edit: The retail CPU I picked up for myself can hit 4.6ghz at 1.115-1.15 volts, and 4.9 on 4 cores (but that may not be stable). This is ~~ 180-190 watts and I consider it "cherry" as far as the limited testing I've done, but far from a unicorn.

2 Likes

Aha, so that's what https://twitter.com/HardwareUnboxed is all about. Ouch. So what's next, for intel, if that is indeed the case? Coming up with a new socket?

I just saw this. What's the point of Intel anymore? Overclocked by 700-900Mhz and you get 5-6 more fps in most games. And that's if you still game on 1080p. Most people with enough money to spend on high-end CPU's probably game on higher resolutions where the CPU is even less important.

Well the effect from overclocking on max frames doesnt really matter all that much.
But what overclocking does, is improving on the minimum frames.
which are the most important frames really.
Max frame rates doesnt really tell all the much about how well a gameplay experiance is.
Its the drops in frames in certain circumstances that counts allot.

They do have the bigger LGA3467 for server parts, but server parts aren't typically meant to be overclocked, and the higher pin count doesn't mean there's more power delivery pins.

1 Like

And if the pricing was similar I'd say "sure, bring on Intel". But it's not, so Ryzen is still a much better choice.

2 Likes