555 - Triple upcoming AMD mid-rangers

[williamcll]

A new mid range motherboard chipset as well as a non-hyperthread Ryzen chip is going to be released soon. The Ryzen 5 3500/3500X (NOT 3500U) with only six cores, directly competing with the intel i5-9400F, the B550 mid ranger motherboard as well as the RX 5500 Graphics card.

(Left: specifications, note the lack of hyperthreading as well as 32MB L3 cache Right: Frame rates, rightmost is world of warcraft)

 

(99th percentile render times in milliseconds)

Quote

AMD will soon be introducing more budget options in their Ryzen 3000 CPU lineup with the arrival of Ryzen 5 3500X and Ryzen 5 3500. Both processors have their specifications and expected prices revealed by Thai media outlet, Extreme IT and look to be very competitive against Intel’s sub $150 US 9th Gen processors.

 

AMD is preparing two variants for the sub-$150 US market, the Ryzen 5 3500X & the Ryzen 5 3500. Both processors feature 6 cores and 6 threads which means that there won’t be multi-threading support on either chip, however, given their price point and the market they are aiming, a cut-down core configuration is expected. As such, these would be the first Ryzen 3000 parts with no multi-threading support.

tarting off with the AMD Ryzen 5 3500X, we are looking at a base clock of 3.6 GHz and a boost clock of 4.1 GHz. The processor features 32 MB of cache, a 65W TDP and a price close to $150 US. It is highly liked that AMD would choose a $149.99 US price for this specific chip.

 

In official slides posted by BullsLab, the chip comes with 24 PCIe Gen 4 lanes, support for 3200 MHz memory and we also get to take a look at some performance benchmarks where the chip is highly competitive versus the Core i5-9400F. The 32 MB Game Cache is also said to deliver much better latencies in a wide variety of esports titles so that is something to consider for gamers on a budget when having to select between an AM4 and the LGA 1151 platforms. The Ryzen 5 3500X is expected to hit retail outlets on 23rd September.

 

The Ryzen 5 3500, on the other hand, has pretty much the same specifications and even the same clock speeds of 3.6 GHz base and 4.1 GHz boost. The difference is that it features 16 MB of cache versus 32 MB on the Ryzen 5 3600X. It would result in a small performance penalty but we are also looking at a lower price point. The chip is said to retail at an expected price point of 4190 Thai Bhat or around $140 US. Given the specs, it is likely that the final prices for the Ryzen 5 3500X would be $149.99 US while the Ryzen 5 3500X could retail for $129.99 US.

Chinese retailers have also listed the Ryzen 5 3500X for 1099 Yuan ($154 US). The Ryzen CPU obviously rocks a more modern 7nm architecture so it will be able to edge out the Intel-based parts with relative ease in most cases at a lower price point. With the launch of these CPUs, AMD would further increase their desktop market share as they aim the mainstream budget tier of the PC market.

Furthermore, an unboxing video of the 3500X was uploaded to BiliBili recently but was taken down, further screenshots are below:

Furthermore, news about the B550 chipset has also been revealed.

Quote

HP has revealed the Pavilion Desktop TP01-0006ng and Desktop M01-F0017ng -- both of which seemingly use AMD's upcoming B550 chipset. The listings also include the first mention we've seen of a new AMD gaming graphics card, the AMD Radeon RX 5300 XT.

The B550 chipset, which is refered to as "AMD Promontory B550A" in both HP documents, will serve as the mid-range chipset for AMD's Ryzen 3000-series processors. ASMedia is responsible for producing the B550 chipset for AMD. The B550 chipset is rumored to lack PCIe 4.0 support, which should help drive down manufacturing costs, however, there isn't enough evidence to confirm that. It's possible that the B550 chipset could use PCIe 4.0 on the first PCIe x16 slot and primary M.2 port. We'll just have to wait to find out.

German retailer Alternate has listed the Pavilion Desktop TP01-0006ng and Desktop M01-F0017ng with arrival dates of October 12 and October 8, respectively. It would seem that the B550 chipset could debut in next month, at least through OEM systems.

But for PC builders, at this point, it's uncertain whether B550 motherboards are even ready yet. In June, DigiTimes reported that ASMedia would ship out B550 orders in the fourth quarter of this year. If that's correct, it's unlikely that motherboard vendors will launch B550-based products in October.

 

 

 

 

Lastly, leaked benchmarks of the RX 5500 has been found, showing that it has a performance similar to the RX 570, additionally, its related navi numbers, including information about the RX 5300 XT have been registration on multiple sites: note that the leak came out after the below news so there is some discrepancies.

Quote

Trivial and urgent. That’s likely not how AMD would want its upcoming Navi 12 GPUs to be referenced, but that’s how its open source guru, Marek Olsak, has termed the addition of the Navi PCI ID to the Mesa 3D Graphics Library in a recent commit. Trivial, presumably because adding the little bit of extra code of Navi 12’s PCI ID doesn’t take a lot of effort, but what of the ‘urgent’ tag? Are we looking at the very imminent arrival of the AMD Navi 12 graphics cards?

It looks like both the Navi 14 and Navi 12 GPUs are going to be arriving before October 15 if this Mesa commit is anything to go by. The trivial and urgent Navi 12 PCI ID add was dropped on September 17 specifically for Mesa 19.2, which only just had its final release candidate launch the following day. Getting in the day before release could be seen as somewhat urgent… Olsak has already dropped Navi 14 support into 19.2 in a little less urgent fashion.

The next Mesa 3D Graphics Library release – 19.3.0-rc1 – isn’t scheduled until October 15 which kinda suggests that AMD’s open source crew wanted to get support into the 19.2 library preceding it, as compatible GPUs would presumably be available before version 19.3 drops.

 

And if the Navi 12 RX 5500 can do that, then it’s going to cause some problems for Nvidia too – it’s no surprise there’s going to have to be a GTX 1650 Ti, because the current GTX 1650 can’t even keep up with the ol’ RX 570 right now. With the GTX 1650 Ti rumoured for launch on October 22 that would give Nvidia time to tune the last throw of the Turing dice this year to deliver against the mainstream Navi onslaught.

And all of this means good things for us PC gamers. The mainstream, sub-$200 graphics card market is the fertile ground from which most of us will pluck our next GPU. And with both AMD and Nvidia genuinely competing for our money by making affordable, high-performance cards we’re all going to be the winners. Yay.

 

Thoughts: I guess even with TSMC's delays all the bad bins are still producing on schedule, Intel really needs to get their game up or else they are going to lose even more customers. The B550 would make a good christmas gift for a gaming PC build.

 

Sources:

Ryzen 5 information:

https://wccftech.com/amd-ryzen-5-3500x-3500-6-core-cpu-benchmarks-prices-specs-leak/
https://item.jd.com/57161665338.html

https://www.tomshardware.com/news/amd-ryzen-5-3500-matisse-zen-2,40227.html

https://www.facebook.com/ExtremeITReview/posts/2379639418810374/

https://twitter.com/BullsLab/status/1174722474405089280

 

B550 motherboard chipset information:

https://www.tomshardware.com/news/amd-b550-chipset-radeon-rx-5300-xt-gpu,40457.html

https://www.digitimes.com/news/a20190614PD212.html

 

RX 5500 graphics card information:

https://gfxbench.com/device.jsp?benchmark=gfx50&os=Windows&api=gl&cpu-arch=x86&hwtype=dGPU&hwname=AMD Radeon RX 5500&did=77731162&D=AMD Radeon RX 5500

https://www.expreview.com/70646.html

https://cgit.freedesktop.org/mesa/mesa/commit/?id=83f195414a2e89bd9f549dacc04365f67e5bd110

https://www.pcgamesn.com/amd/navi-12-navi-14-rx-5600-rx-5500-october-15-launch

 

[givingtnt]

Neat

[PlayStation 2]

...ya know, I was planning on building my younger brother a PC for his graduation gift. That 3500X is looking mighty neat.

[xAcid9]

My local PC builders really love to recommend 9400f for "budget" build, i reckon they will still recommend 9400f.

[williamcll]

I can't remove the final photo for some reason, it just keeps coming back everytime I edit the post.

[dizmo]

Soooo $10 more for ever-so-slightly worse performance?
Makes sense.

[Shreyas1]

CPU looks neat, though still wondering when AMD will challenge NVIDIA in the higher end

[williamcll]

Just now, Shreyas1 said:

CPU looks neat, though still wondering when AMD will challenge NVIDIA in the higher end

Wait for 5900 XT or Vega VII pro duo

[williamcll]

56 minutes ago, huilun02 said:

It would be nice if AMD could go ham and make a 8c8t with 6Ghz base clock or something stupid. So they can just end Intel.

Remember the last time they did one with 5Ghz?

[Guest]

6C/6T no SMT???

Spoiler

 

[GoldenLag]

Welp, so a worse CPU choice over the r5 2600 and the 9400f.........

 

If it was like 120$, maybe? Essentially this should be an r3.

 

Spoiler

 

[porina]

I had posted this in another non-news thread previously. There was a link to what looks like a Chinese store, where the 3500X was listed around 2/3 the price of a 3600. I think that might be a better indication of positioning than trying a direct currency conversion.

 

I'm thinking this CPU could be a good choice for dedicated crunchers for workloads that don't benefit from SMT. At that point, the other thread degenerated into getting told why we need SMT.

[williamcll]

39 minutes ago, GoldenLag said:

Welp, so a worse CPU choice over the r5 2600 and the 9400f.........

 

If it was like 120$, maybe? Essentially this should be an r3.

 

  Reveal hidden contents

 

That's whatever quad core that is going to come out later.

[GoldenLag]

7 minutes ago, williamcll said:

That's whatever quad core that is going to come out later.

Yes? But the 3500 is also just pointless cashgrab. Especially the 3500x. 

 

If the R3 4c has SMT ill be laughing. 

 

Because the 4c/8t will probably have less stutter than the 6c/6t if 7700k vs 8600k is anything to go by.

[Guest]

[WereCatf]

I fail to see the point.

[porina]

1 hour ago, GoldenLag said:

If the R3 4c has SMT ill be laughing. 

 

Because the 4c/8t will probably have less stutter than the 6c/6t if 7700k vs 8600k is anything to go by.

Got an example of that? It isn't something I looked into previously for gaming, but for most things compute 6c6t will be the better choice than 4c8t assuming similar clocks and other factors. For 4c8t to take the lead it would have to scale extremely well with SMT. Even Cinebench R15, which scales pretty well with SMT, should favour 6c6t. (I don't have data for Zen2 yet, but this is certainly applicable to Intel CPUs and Zen/Zen+)

[GoldenLag]

2 minutes ago, porina said:

Got an example of that? It isn't something I looked into previously for gaming, but for most things compute 6c6t will be the better choice than 4c8t assuming similar clocks and other factors. For 4c8t to take the lead it would have to scale extremely well with SMT. Even Cinebench R15, which scales pretty well with SMT, should favour 6c6t.

battlefield or AC: oddessey are the best examples of it. where the 4c/4t are dropping in 1% lows and 6c/6t are suffering to a lesser degree. 4c/8t wont reach the same average clockspeed, but it doesnt got the drops the 6c/6t gets

[samcool55]

Umm, isn't this DOA because of the 2600x? There is like a 5/10 dollar difference (if the price is 150-ish) and I doubt the IPC gain can pull ahead against SMT.

[porina]

15 minutes ago, samcool55 said:

Umm, isn't this DOA because of the 2600x? There is like a 5/10 dollar difference (if the price is 150-ish) and I doubt the IPC gain can pull ahead against SMT.

I'd take a 3500 over any 2000 series any day, for my uses at least. Zen 2 has much better FPU and cache, it obsoletes older generations as far as I'm concerned.

 

Actually, I don't know what the IPC uplift is for Cinebench going to Zen 2 (with and without SMT), but the given figure from memory was around 18%. The benefit from SMT in Cinebench is of the ball park 30%. So while it doesn't completely close the gap, it takes a large chunk out of it. For not so well scaling with SMT applications (most things) it would be closer still.

[Doobeedoo]

Well nice SKU very good options now really, I take it it will be popular. 

[The_Prycer]

6 hours ago, williamcll said:

Remember the last time they did one with 5Ghz?

Yes, I'm using one in my home rig right now.

 

I still want one "9590" for every release cycle where AMD just says "to hell with power efficiency or anything resembling rational thought, have this 8/8 CPU with an obscene base clock that will require you to learn about water cooling or invest in a submersible case because it's AWESOME"

 

[spartaman64]

the main game i play (bdo) is so poorly optimized that you actually see better performance when you disable hyperthreading and SMT. ive heard of people getting 35 fps more when they disable hyperthreading so maybe i want to get the 3500

[5x5]

10 hours ago, Shreyas1 said:

CPU looks neat, though still wondering when AMD will challenge NVIDIA in the higher end

That's not lucrative. A 5700 or 2060S makes several times more money than a 2080A or 2080 Ti. The high end market is by far the smallest and the low end 1030-1660 is by far the largesr

[ThePD]

16 hours ago, williamcll said:

Wait for 5900 XT or Vega VII pro duo

Think AMD registered the 5950XT and 5950 SKU's as well.