News Amazon Italy Reveals AMD Ryzen 9 3900XT, Ryzen 5 3600XT Pricing and Release Date

joeblowsmynose

Distinguished
Yep it seems that AMD is learning from Intel, less than 1% increase in performance and 25% increase in price

Yeah that doesn't make sense ... if the only difference is +100mhz boost and say the rumoured +200mhz base, that will be nice and all, but not if the prices are 20-25% higher ... Maybe there's more up AMDs sleeve with these chips (naybe they OC well)? or maybe the prices won't actually stay at that for any length of time.

I still am waiting out getting a 3900x - I thought the XT might be a nice option, but not for that premium, if it turns out accurate. If the XT can (stably) all core OC to 4.4 without being a nuclear furnace, then maybe ...
 

InvalidError

Titan
Moderator
Usually, a "refresh" means that the chip has changed in some significant way such as a die re-spin or refined fab process. AFAIK, these are the same old Zen 2 dies, just more finely binned so AMD can cash in on chips guaranteed to be a whole 2-5% faster stock.
 

joeblowsmynose

Distinguished
Usually, a "refresh" means that the chip has changed in some significant way such as a die re-spin or refined fab process. AFAIK, these are the same old Zen 2 dies, just more finely binned so AMD can cash in on chips guaranteed to be a whole 2-5% faster stock.

That's what it appears - they just took some CCXs that we're binned for 3950x and plugged them into other skus.
 
You do know that EU public prices always include taxes, right? Around 20% depending on the country? As such the actual US prices are closer to €475 before taxes, making that CPU closer to $532. But, since US companies usually shaft Euro customers on exchange rates, you can expect this chip to come out at $475 on your side of the pond.
 
As the article states, that may very well just be placeholder pricing until the official prices get announced. Considering an 8-core, 16-thread 3700X can now be had for just $275 at most major online stores, I don't think we'll see a 3600XT with two fewer cores and a smaller stock cooler for $45 more.

We also don't know what multi-core boost clocks will be like though. It's possible that the multi-core boost could remain closer to the single-core boost, like what we saw with the Zen+ processors compared to their predecessors.
 

spongiemaster

Admirable
Dec 12, 2019
2,273
1,277
7,560
You do know that EU public prices always include taxes, right? Around 20% depending on the country? As such the actual US prices are closer to €475 before taxes, making that CPU closer to $532. But, since US companies usually shaft Euro customers on exchange rates, you can expect this chip to come out at $475 on your side of the pond.
The price comparison was between the 2 prices on the Italian site which doesn't involve the exchange rate to US dollars. The 3900XT was listed at €569.69 while the 3900X currently lists for €455.10. That's a 25% price increase comparing apples to apples.
 
pricing for a mere 100mhz should be near same as current ones.

Doubt the OC potential is any better given how the limit seems to be a hardlimit with the architecture.

why pay more when you can OC and hit same while spending less?
 

InvalidError

Titan
Moderator
why pay more when you can OC and hit same while spending less?
Many people don't want to bother with OCing and the likelihood of getting screwed over by the silicon lottery, especially on Zen 2 where manual OC is often worse than stock in workloads that benefit more from automatic few-cores boost. Also, since AMD likely bins dies for those new SKUs, lower-end parts may very well end up with demonstrably worse silicon on average than before.
 

joeblowsmynose

Distinguished
Many people don't want to bother with OCing and the likelihood of getting screwed over by the silicon lottery, especially on Zen 2 where manual OC is often worse than stock in workloads that benefit more from automatic few-cores boost. Also, since AMD likely bins dies for those new SKUs, lower-end parts may very well end up with demonstrably worse silicon on average than before.

And how would you demonstrate "demonstrably"? AMD chips starting to fail at stock? Or ... Chips that don't OC anywhere near their boost clocks in most cases, just like now, except not doing that ... more??

Those are about your only two options there, and the 2nd one is moot, and the 1st one isn't reasonable.

It would be far more likely that AMD is running into an excess of better silicon now that the production has had a full year to mature, so might as well capitalize on that by putting the CCXs that are fully capable of being 3950Xs, but the demand for that chip <> the quantity of good silicon, therefore, use those CCXs to create new lower core count skus -- the more popular ones ...

I don't buy the idea that AMD is only doing this to counter Comet lake .... they don't really need to counter with zen3 right around the corner, so to me, the idea of just trying to utilize the best silicon in the way that will make them sell more of that high quality silicon, and maybe get a little better premium ... that makes the most sense to me.

I still don't think the price should exceed the launch prices for their original counterparts, that would be a fail, if they did, in my opinion.
 

InvalidError

Titan
Moderator
And how would you demonstrate "demonstrably"?
Gather a large sample size over time and see how people's results change. In the past, we used to have Silicon Lottery doing this but due to how Zen 2 had practically no meaningful overclocking headroom for SL to make a profit on, they quit bothering.

As for AMD having an excess of 3950/TR/EPYC-worthy parts, AMD can burn those on 3100 and 3300X too as impetus to up-sell people from the 3600/3600X to the 3600XT instead of letting those percolate through the product range and improve overclocking margins on existing SKUs. AMD has a vested interest in reducing the amount of overlap between neighboring pigeonholes in its product stack.
 

spongiemaster

Admirable
Dec 12, 2019
2,273
1,277
7,560
I don't buy the idea that AMD is only doing this to counter Comet lake .... they don't really need to counter with zen3 right around the corner, so to me, the idea of just trying to utilize the best silicon in the way that will make them sell more of that high quality silicon, and maybe get a little better premium ... that makes the most sense to me.

Or Zen 3 isn't right around the corner. These unannounced CPU's wouldn't be coming to market if Zen3 was getting released within the next quarter. You wouldn't want to risk someone upgrading just a CPU, when you could get them to buy a more expensive cpu and a motherboard for Zen3 due to Zen3's fractured backwards compatibility. This year could mean December announcement, volume shipping next year.
 

joeblowsmynose

Distinguished
Or Zen 3 isn't right around the corner. These unannounced CPU's wouldn't be coming to market if Zen3 was getting released within the next quarter. You wouldn't want to risk someone upgrading just a CPU, when you could get them to buy a more expensive cpu and a motherboard for Zen3 due to Zen3's fractured backwards compatibility. This year could mean December announcement, volume shipping next year.

Zen2 still competes perfectly well against comet lake. If you do more than gaming, and even if you only game but don't have a 2080ti on low game settings, then Zen2 is still being the recommended as the preferred part for builders by all the most respected tech experts. Comet lake didn't really change much at all.

AMD has very recently ... and again, re-iterated, that Zen3 will be launching in Q4 ... the recent delay rumours were based on some silly notion that AMD was delaying Zen3 to move it to 5nm --- AMD has clarified that this is not true at all - Zen3 is on 7nm, the original roadmap has never changed, except AMD now has RDNA2 for mobile phones (samsung) to make -- those will likely be going into 5nm - so that's where that silly rumour came from.

Q4 begins in three months. Intel won't be releasing anything between now and then so nothing changes in the dynamic between the two company's parts between now and then.

A CPU purchased today has more value than one purchased in the future ... If AMD wasn't trying to ensure longevity for the AM4 platform they would have just done it like Intel all these years ... they didn't. You are assuming that they feel that was a mistake and will be stealing future sales by continuing to support the existing products still on the market. Nope. If they thought that way AM4 wouldn't have been what it is. That's how Intel thinks, so I understand applying that expectation. There's huge mindshare value in properly supporting the lobgevity of a platform. That will translate into $$ on its own merit, perhaps less obviously, but intangible benefits are very real and something every business and marketing dept must consider, especially when trying to build a brand over time.
 
Last edited:

joeblowsmynose

Distinguished
... but due to how Zen 2 had practically no meaningful overclocking headroom for SL to make a profit on, they quit bothering.

Yup, that was my point, and why its moot. Nobody has any expectation for being able to hit high all core frequencies anyway.

As for AMD having an excess of 3950/TR/EPYC-worthy parts, AMD can burn those on 3100 and 3300X too as impetus to up-sell people from the 3600/3600X to the 3600XT instead of letting those percolate through the product range and improve overclocking margins on existing SKUs. AMD has a vested interest in reducing the amount of overlap between neighboring pigeonholes in its product stack.

Yeah that might make sense ... the 3100 and 3300x actually do appear to be higher quality silicon than their earlier brethren ... But they would have no reason to make those exclusive of other skus.
 
Last edited:

InvalidError

Titan
Moderator
If AMD wasn't trying to ensure longevity for the AM4 platform they would have just done it like Intel all these years ... they didn't.
AMD admitted in its Zen 2 launch last year that getting it to work on AM4 proved more difficult than it wanted it to be - had to "get creative" with substrate and layout. I wouldn't be too surprised if at least part of the reason why AMD wanted to ditch Zen 3 compatibility with pre-500 boards is because Zen 3 won't have some of the bodges that Zen 2 needed to make it work with quirkier 300/400-series boards similar to how it forced a move to sTRX4 on the ThreadRipper side of things.

I doubt AMD will commit to more than two years of support ahead of time for AM5. Design next-gen CPUs, evaluate any compromises that may arise from maintaining compatibility with the current socket and chipsets, announce one more generation of support if there are no substantial hurdles, announce a new socket to address deficiencies otherwise.

Personally, I don't use new CPUs on old motherboards so I couldn't care less about forward-compatibility and would much prefer putting that effort towards rock-solid single/dual-gen boards instead.
 

joeblowsmynose

Distinguished
AMD admitted in its Zen 2 launch last year that getting it to work on AM4 proved more difficult than it wanted it to be - had to "get creative" with substrate and layout. I wouldn't be too surprised if at least part of the reason why AMD wanted to ditch Zen 3 compatibility with pre-500 boards is because Zen 3 won't have some of the bodges that Zen 2 needed to make it work with quirkier 300/400-series boards similar to how it forced a move to sTRX4 on the ThreadRipper side of things.

I doubt AMD will commit to more than two years of support ahead of time for AM5. Design next-gen CPUs, evaluate any compromises that may arise from maintaining compatibility with the current socket and chipsets, announce one more generation of support if there are no substantial hurdles, announce a new socket to address deficiencies otherwise.

Personally, I don't use new CPUs on old motherboards so I couldn't care less about forward-compatibility and would much prefer putting that effort towards rock-solid single/dual-gen boards instead.

Even if they didn't back peddle on the "no zen3 support on old boards" decision, it still would have been four generations (if you count early AM4 APUs on bulldozer) on AM4 and support to 2020 -- which was their initial commitment. So it would have been mostly easy sailing to cover all that ... aside from the "too small bios" issue on MSI boards (which was sort of fixed with a crappy bios version), I think that was the main issue with the zen2 compatibility. But that was board makers issue.

I think the fact that they got PCIe 3 / 4 in there and kept compatibility is applaudable - should be seen as a marketing positive, perhaps.
 
Last edited:

spongiemaster

Admirable
Dec 12, 2019
2,273
1,277
7,560
AMD has very recently ... and again, re-iterated, that Zen3 will be launching in Q4 ... the recent delay rumours were based on some silly notion that AMD was delaying Zen3 to move it to 5nm --- AMD has clarified that this is not true at all - Zen3 is on 7nm, the original roadmap has never changed, except AMD now has RDNA2 for mobile phones (samsung) to make -- those will likely be going into 5nm - so that's where that silly rumour came from.

Q4 begins in three months. Intel won't be releasing anything between now and then so nothing changes in the dynamic between the two company's parts between now and then.

AMD has never said Zen3 is launching in Q4 of 2020 and nothing in your link has any information from AMD for a launch window.

As of last month, an actual quote from Lisa Su was:

"on track to launch our next-generation Zen 3 CPUs and RDNA 2 GPUs in late 2020."

If October was an option, the quote would have been Q4, not late 2020. And unfortunately, as is pretty much industry standard now, launch does not mean product availability. That usually comes a month of more after the launch. Like I said previously, Zen3 could hit shelves in December, with broader availability in early 2021. Zen3 is not right around the corner, it could well be another 6 months which gives a reasonable window to allow for the release of these tweaked 3000 chips. I would expect Intel to paper launch Rocket Lake in the same time frame as Zen 3 with no availability until early 2021.
 

Makaveli

Splendid
Yeah that doesn't make sense ... if the only difference is +100mhz boost and say the rumoured +200mhz base, that will be nice and all, but not if the prices are 20-25% higher ... Maybe there's more up AMDs sleeve with these chips (naybe they OC well)? or maybe the prices won't actually stay at that for any length of time.

I still am waiting out getting a 3900x - I thought the XT might be a nice option, but not for that premium, if it turns out accurate. If the XT can (stably) all core OC to 4.4 without being a nuclear furnace, then maybe ...

The mystery with these chips is still what FCLK will they support if the rumor'd 2000 is correct plus the 200+ base and 100+ boost and better binned silicon it maybe provide quite a boost when paired with DDR4 4000 Memory.
 

Nightseer

Prominent
Jun 12, 2020
17
7
515
I would like to just warn everyone about European prices, unlike USA prices, all prices in Europe include VAT. Which amounts to about 20% price increase. So before converting in USD, remove those 20% of VAT and you will get price you would see in USA. Just saying, since some store do 1:1 conversion with 20% tax included. Just saying.
 

P1nky

Distinguished
May 26, 2009
60
25
18,560
the Ryzen 9 3900XT and Ryzen 5 3600XT apparently cost €569.69 ($640) and €284.84 ($320),

Wrong. That's not how you do UE currency conversion. First you have to remove the VAT and then convert to USD. Italy has 22% VAT, so the prices for 3900XT should be $500 and 3600XT should be $250.