AMD Bristol/Stoney Ridge Thread

Page 34 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

dark zero

Platinum Member
Jun 2, 2015
2,507
8
61
I see what they are doing...
Zen Lite is Jaguar with Steroids and L3 cache

So it will have greater ST performance than BD, but MT wise it will need more power.

They uses that in order to reduce consumption and to make it way smaller.... But what is the intention?

Taking down Atom?
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
I see what they are doing...
Zen Lite is Jaguar with Steroids and L3 cache

So it will have greater ST performance than BD, but MT wise it will need more power.
Zen-Lite should be Bulldozer-ish with the focus of not being a "Premium" High Performance core. To save design costs; the Zen front-end stays; 64 KB L1i / NNBP / etc. Depending on if they plan to completely restart it. 4-wide decode to two 2-wide clusters like Bulldozer, or two 2-wide decode to two 2-wide clusters like Excavator. Actual core-wise it should be more Jaguar than Excavator. Excavator has the FPU not in the core cluster, while Jaguar and Zen has the FPU in the core cluster. The main change between Zen/Jaguar to Zen-Lite is the fusion of the Integer Rename and Floating Point Rename; Mapper and Non-scheduling Queue becomes one unit. This in turn reduces the wiring from two units to one unit. Primary benefit is the retire queue gets to remain huge with this. It may also resolve dependencies across the LDC faster than Zen/Jaguar.

The CCX stays as the change from SMT to CMT is not visual beyond the L2 cache. It looks like a standard SMT2 core to the L3 caches control and control/data fabrics, and other things.

MCMT and Cluster-based Multithreading has a EPI bonus in every early design. EPI focus means that the architecture can maybe utilize the extended voltage range that FDSOI supports. 22FDX is capable of a larger Vdd range than with 22PDSOI. Extended voltage range means extended frequency range... which means >5 GHz single module boosts can be done. So, add FDSOI self heating and AMD's Ti-states. Ti-states are temperature inversions, it is where the more hot a transistor is the better the device performs. A FDSOI transistor being a hot ah heck has a larger Ti-state range than FinFET transistors do.
They uses that in order to reduce consumption and to make it way smaller.... But what is the intention?

Taking down Atom?
It is a more cost-effective alternative than Banded Kestrel/Raven Ridge-L. Since, it is replacing those, each APU die including the CPU versions have a single 64-bit DDR4/LPDDR4(X) multi-model controller/PHY. What that means is AMD can produce a "Value Threadripper" that competes with Denverton(Goldmont)/Jacobsville(Tremont). These chips would utilize the same standard AM4 socket.

Dual Module CPU + 8 CU GPU => <$83 APU -> 1x 64-bit DDR4
Quad Module CPU => <$75 CPU -> 1x 64-bit DDR4
2x Quad Module CPU => <$170 CPU -> 2x 64-bit DDR4
... for example.

If we use the historic Athlon AM1 for reference though...

Dual Module CPU + 8 CU GPU => <$54 APU -> 1x 64-bit DDR4
Quad Module CPU => <$51 CPU -> 1x 64-bit DDR4
2x Quad Module CPU => <$97 CPU -> 2x 64-bit DDR4

So, instead of a small 2 Zen core/ 3 GCN die that costs $xx. AMD sells a medium sized 4 Zen-lite/ 8 GCN die that costs $xx. $xx being the same between the two, etc. The larger SKU has more utility for the mainstream than the smaller SKU.
 
Last edited:

NTMBK

Diamond Member
Nov 14, 2011
8,250
227
126
Mainstream is wider overall in the market. Expanding the mainstream profit margin is all about having a better product than the competitor. Mainstream can be sold into markets that naturally have low spending budgets. A product purchased is better than a product not purchased because it was too expensive.

Mainstream also supports the weight of the premium market. Testing and compiling as general performance goes to the mainstream product first. Then, to the higher performance product. So, it is important to have those dev kits be cheap!
The budget end of the market is the section that's being eaten alive by smartphones, and the existing install base. People don't need to buy craptops and budget desktops to go on Facebook and check emails any more- they can either use their 5 year old system that works fine, or just use their phone. How much really is there that's worth addressing below Raven Ridge? I could maybe see the justification for making a specific die for that market with e.g. two Zen cores, half the GPU shaders, and a single channel of memory (basically a dedicated die for the Ryzen 3 2200U, instead of just using salvaged Raven Ridge dies). But coming up with an entire new CPU architecture? No way is there enough money in that market to justify it, not for a company of AMD's size.
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
But coming up with an entire new CPU architecture?
Neither new or from scratch.
Simultaneous Multithreading to Cluster-based Multithreading process;
1. Reuse
2. Resolve
3. Test
4. Profit
How much really is there that's worth addressing below Raven Ridge?
While it is below Raven Ridge, it is mostly across.
4 cores with High IPC and 11 CUs. -> Premium/Performance
4 cores with Low EPI and 8 CUs. -> Mainstream/Value

The market extension is for those with Trinity/Richland/Kaveri/Godavari/Carrizo/Bristol Ridge. While also being for Brazos(2.0/T), Kabini(Temash), Beema(Mullins), Carrizo-L, and Stoney Ridge. All of these flow into the 4-core/8CU design, not the Raven Ridge2/-L 2-core/3CU design. A four core with 8 compute units APU at 4.5W~10W is more sell worthy than something that is two cores with 3 compute units APU in the same TDP range.

-> 22FDX
-> Same Zen tiles/instances to form CMT instead of SMT. (SMT2_F4 to CMT2_F2_C2 is cheap crossover)
-> Doesn't utilize Malta fab, but utilizes Dresden/Chengdu. Open for low-cost third party semi-custom partners in europe(France-> Atari S. A.) and china(MediaTek/Hygon Consumer), etc.
-> Bristol Ridge performance at lower TDP and higher at same TDP. All the while being more cost-effective alternative to 2C/3CU 14LPP/12LP.
-> Fast dev cycle means AV1 HW decoder can be included by a Q3~Q4 2018 launch.
 
Last edited:

NTMBK

Diamond Member
Nov 14, 2011
8,250
227
126
Neither new or from scratch.
Simultaneous Multithreading to Cluster-based Multithreading process;
1. Reuse
2. Resolve
3. Test
4. ProfitWhile it is below Raven Ridge, it is mostly across.
4 cores with High IPC and 11 CUs. -> Premium/Performance
4 cores with Low EPI and 8 CUs. -> Mainstream/Value

The market extension is for those with Trinity/Richland/Kaveri/Godavari/Carrizo/Bristol Ridge. While also being for Brazos(2.0/T), Kabini(Temash), Beema(Mullins), Carrizo-L, and Stoney Ridge. All of these flow into the 4-core/8CU design, not the Raven Ridge2/-L 2-core/3CU design. A four core with 8 compute units APU at 4.5W~10W is more sell worthy than something that is two cores with 3 compute units APU in the same TDP range.

-> 22FDX
-> Same Zen tiles/instances to form CMT instead of SMT. (SMT2_F4 to CMT2_F2_C2 is cheap crossover)
-> Doesn't utilize Malta fab, but utilizes Dresden/Chengdu. Open for low-cost third party semi-custom partners in europe(France-> Atari S. A.) and china(MediaTek/Hygon Consumer), etc.
-> Bristol Ridge performance at lower TDP and higher at same TDP. All the while being more cost-effective alternative to 2C/3CU 14LPP/12LP.
-> Fast dev cycle means AV1 HW decoder can be included by a Q3~Q4 2018 launch.
Why would anyone with an existing Kaveri or Carrizo laptop want something that low end? You might as well just stick with the decent laptop you have. Minimal acceptable upgrade would be Raven Ridge level. It's the natural replacement, with a smaller die size than Llano, Trinity, Kaveri or Carrizo.
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
Why would anyone with an existing Kaveri or Carrizo laptop want something that low end?
In no way is it actually low-end. It is at least middle of mainstream, where Raven Ridge is lower part of performance.

Low EPI means high frequency. (ex: 2x frequency at same power or 1x frequency at half // power scales better so 1x frequency at 1/10th)
- Add 22FDX to the scene it now has a ultra-wide voltage-frequency range to use.
- Add body biasing.
- Throw in a fast and efficient IVR.
- With an eager resolve of Zen to 2xZen-lite CMT(2 ALU/2 AGU/2 FPU: no new units). (Essentially, super-Jaguar or lite-Zen.)

At its extreme it is a worthy x86 smartphone SoC. If Nvidia can put a modem on a 28nm HPM SoC, then AMD via semi-custom version can put a modem on a 22FDX SoC.
You might as well just stick with the decent laptop you have. Minimal acceptable upgrade would be Raven Ridge level. It's the natural replacement, with a smaller die size than Llano, Trinity, Kaveri or Carrizo.
I have Bristol Ridge, minimum acceptable upgrade is the Pentium N5005. VP9 encode is total worth, but AV1 is around the corner. One semi is doing 8K120 AV1.. so attach that to a HDMI 2.1 48 Gbps Tx. That chip that gets that has got everything covered for a llloooooongg time.

My calcs estimate the die area for 22FDX would be at max 190 mm squared and at min 170 mm squared. (GPU-shrink, CPU-shrink, PHY-shrink, MEM-IO shrink, etc.)

Q4 2018;
$11 per die for Bristol Ridge(28HPA w/ SHP transistors)
$8 per die for the 22FDX(+)(For UHP transistors)
$16.5 per die for Raven Ridge(14nm+ transistors)

Q4 2020;
$11.8 per die for 12FDX(+) w/ LPDDR5/DDR5 bi-model controller/phy and 8 core/16 CU.
 

Abwx

Diamond Member
Apr 2, 2011
8,805
132
126
(ex: 2x frequency at same power or 1x frequency at half // power scales better so 1x frequency at 1/10th)
Total non sense, like the rest of your wild speculations that never materialize one way or another...

You realize that with GF s 14nm LPP clocking something at 2X the frequency require almost 6X the power, and that their FDX SOI process is no better in this respect...?

I would suggest some homework about transistors physics and general theory, you would spare yourself quite a load of uselessness in mounting scenariis that have no chance to occur with the current physics laws...

https://en.wikipedia.org/wiki/Field-effect_transistor
https://en.wikipedia.org/wiki/MOSFET
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
You realize that with GF s 14nm LPP clocking something at 2X the frequency require almost 6X the power, and that their FDX SOI process is no better in this respect...?
I don't know what you are talking about but FDSOI has a magical pixie dust across the wafer called Ultra-Thin Buried Oxide(25nm to 8nm of SiO2). This magical pixie dust allows for a diode to apply a voltage on on the bottom of a P-well or N-well to get certain types of bias; Forward or Reverse. Forward allows for trimming which reduces Pdyn and reverse allows for buffing which reduces Pstat. Which in turn reduces Ptot/Pavg that in turn allows for higher frequency at same Ptot/Pavg/PTDP.

This isn't a same node transition though;
- It isn't a 22FDX device going to an improved 22FDX device for that 2X frequency boost.
- It is a 28nm Bulk device going to a 22FDX(+) device for that 2X frequency boost.

There is also other factors as well;
- Different architecture (fan-out/wire length/transistors)
- Different system architecture (FBB and RBB)
- Different IVRs (microsecond response versus millisecond response)
- Different Vmax/Vmin/AVt (higher overall yield)
- etc
Allows for increased frequency if the energy used is less.

Three 50% power shrinks equals three 30% performance increases which 22FDX+ provides over 28nm HPA. A change in architecture to better fit equates to four 50% power shrinks and four 30% performance increases.

Zen-Lite calcs on 22FDX+;
sub-two gigahertz << sub 4.5 watts of TDP
sub-four gigahertz << sub 15 watts of TDP ~3.y times power increase
sub-six gigahertz << sub 65 watts of TDP ~4.y times power increase for half the frequency.
 
Last edited:

ET

Senior member
Oct 12, 1999
513
26
91
Why would anyone with an existing Kaveri or Carrizo laptop want something that low end?
Longer battery life / smaller form factor, and better GPU performance. AMD laptops still have bad battery life, even with Raven Ridge, and the 15W power envelope brings the Bristol Ridge GPU to around the Intel IGP level. AMD needs something that has reasonable performance and low power. I'm hoping a 7nm Zen APU would finally fill the spot, but if AMD comes up with something else which achieves this and is lower cost, I think it would have a market.

But NostaSeronx is discussing the CPU side only, and I wonder what benefits a GPU gets at 22FDX+ and what AMD might choose to offer there. That's a big part of the equation.

In any case, I'm definitely not planning to make any purchasing decisions based on these rumours. I'll wait to see what turns up, if anything. Still, NostaSeronx's speculations are certainly technically interesting, so I always welcome them.
 

amd6502

Senior member
Apr 21, 2017
457
54
71
Longer battery life / smaller form factor, and better GPU performance. AMD laptops still have bad battery life, even with Raven Ridge, and the 15W power envelope brings the Bristol Ridge GPU to around the Intel IGP level. AMD needs something that has reasonable performance and low power [....]

But NostaSeronx is discussing the CPU side only, and I wonder what benefits a GPU gets at 22FDX+ and what AMD might choose to offer there. That's a big part of the equation.

In any case, I'm definitely not planning to make any purchasing decisions based on these rumours. I'll wait to see what turns up, if anything. Still, NostaSeronx's speculations are certainly technically interesting, so I always welcome them.
I think battery life since BR has been very good for 10W-15W. Often, the issue is that windows and OEMs don't seem to set up a decent battery profile. They should at least limit the battery profile max freq to 99% of max frequency; that way the boost frequencies are disabled and this can extend battery life significantly (more aggressive batt profiles, ex max freq <80%, extends battery life even further.) Wifi and bluetooth setting also have a big effect, as do OEMs choice of memory (voltage and nr of memory sticks---here single channel single stick low wattage memory gets a significant advantage).

Linux (at least for older products like Kaveri) is even worse, idling the iGPU at max frequency (I get well over twice the typical battery life when I boot to Win10 versus when I'm on using linux). I'm not sure about newer stuff like RR, but I'm guessing the state of battery life under linux is unchanged.

For the ~6w APUs, the choice might be pretty lacking (I don't consider Stoney anything but a mediocre stopgap) right now; this would be best addressed by a native 12nm 2c/4t Zen+ with smaller iGPU.

Assuming such a 2c/4t Zen+ APU, what segments could some hypothetical 22FDX product fill? Not too many, and that's going to limit the size of their investment to complete shoestring.

Maybe they can dual purpose an FDX product to a die that produces both a low end discrete GPU (under RX550, ~6CU) and a lower end APU (4 puma+4CU); (or something vaguely similar like a excavator+puma 2c+2c hybrid).

Could Radeon 530 be a case of such dual purposing from Carrizo-L / CZ / BR dies (It has five to six GCN 1.2 units, and runs both GDDR5 and DDR3). Given the RAM shortage and surplus of refurb 2GB DDR3 laptop sticks; this would be a very interesting use of Carrizo.
 
Last edited:

ET

Senior member
Oct 12, 1999
513
26
91
Could Radeon 530 be a case of such dual purposing from Carrizo-L / CZ / BR dies (It has five to six GCN 1.2 units, and runs both GDDR5 and DDR3).
Is the 530 really GCN 1.2? Are there any actual details about it? I thought it was just the same old GCN 1.0 Oland that's been used for all AMD low end mobile chips for a while. Why would AMD replace a 77mm chip with a huge APU die?

(Really, if the 530 is GCN 1.2, it will affect how I recommend it to users of our software, especially if it includes 4K video decode/encode.)

As for the FDX product, I agree that I see no great reason to create one except for using the process. Which may be an okay reason if it does have good performance / power / price and there's demand for custom chips for it.
 
Last edited:

NTMBK

Diamond Member
Nov 14, 2011
8,250
227
126
Could Radeon 530 be a case of such dual purposing from Carrizo-L / CZ / BR dies (It has five to six GCN 1.2 units, and runs both GDDR5 and DDR3). Given the RAM shortage and surplus of refurb 2GB DDR3 laptop sticks; this would be a very interesting use of Carrizo.
The 530 has the exact same spec sheet as an old Oland GPU; same number of ROPs and texture units, same clock speeds, same performance numbers. Given that we've never seen a Carrizo/Bristol Ridge unit supporting GDDR5, it feels more likely that it's an Oland chip and not actually GCN 1.2, though it's hard to find solid evidence. (Also the AMD website lists it as not supporting a whole bunch of modern video decode formats that Bristol Ridge supports, which also rules it out for me.)
 

ET

Senior member
Oct 12, 1999
513
26
91
You know what would really excite me? I'd love for AMD to replace Oland with a 22FDX chip. Even if it's the same number of CU, just have GCN 1.2 at least, with its video block. In terms of die size, the process should compensate for the extra complexity, plus make it lower power and hopefully cheap enough to be attractive.

Probably not too exciting for users, but I hate the endless rebranding and just moving the market a couple of generations forward would be really helpful
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
Lets called it XLP3 from now on. XLP3 maybe started in 2014 as a project from leftover Jaguar and Excavator teams. Essentially, Zen design methodology but with CMT: 2 narrow (1T) cores(XLP3) or 1 wide (2T) core(SHP2). 22FDX physical design started in 2015, before the switch to 22FDX+ in 2H 2017(sometime before September).
//SHP2 = Zen, SHP1 = Bulldozer, XLP1 = Bobcat, XLP2 = Jaguar

Following the XLPx series..
Bobcat(XLP1) -> 4.9 mm squared
Jaguar(XLP2) -> 3.1 mm squared
The shift to 22FDX 8T is the equivalent of going from 28nm 9T to 28nm 7T in area. Going from XV would be ~10.4 mm squared, going up from Jaguar is ~4.7 mm, and going across from Zen would be ~6.3 mm; mean is 7.1 mm squared; 5.9 mm squared for up Jaguar/across Zen. If it is less than this even better...

XLPx isn't known for its clock rate;
Bobcat -> 1.75 GHz (40nm)
Jaguar -> 2.5 GHz (28nm)
Trend-wise the XLP3 core for 15W would be 2.75 GHz to 3.25 GHz. (22FDX+ (equiv:20SHP/20A))

With a condensed frequency and voltage range the complexity is lowered making a density shrink. So, area wise Octo-core on the CPU side is possible if design similarities to Zen is kept.

Area of competition;
Nvidia Denver/Carmel (Dual-core/Octo-core)
Intel Goldmont/Tremont (Quad-core)
VIA WuDaoKou/LuJiaZui (Octo-core)
ARM Cortex A75/Kryo 385 (Quad-core/Octo-core(8 cores in a cluster are supported))

I/O wise everything should be smaller in 22FDX compared to 28HPA. 128-bit LPDDR4/DDR4 (22nm) is smaller than 128-bit DDR3/DDR4 (28nm). Also, 3.2 GHz in LPDDR4/DDR4 should consume about the same power as 2.4 GHz in DDR3/DDR4.

With a 128-bit memory I/O the clock rate for the GPU can be boosted in match with the increase in memory bandwidth. The GPU is more closer to the Vdd_nom than the CPU is which means it gets more benefit of body biasing. More performance at less than half power, overall the device power drops.
 
Last edited:

amd6502

Senior member
Apr 21, 2017
457
54
71
The 530 has the exact same spec sheet as an old Oland GPU; same number of ROPs and texture units, same clock speeds, same performance numbers. Given that we've never seen a Carrizo/Bristol Ridge unit supporting GDDR5, it feels more likely that it's an Oland chip and not actually GCN 1.2, though it's hard to find solid evidence. (Also the AMD website lists it as not supporting a whole bunch of modern video decode formats that Bristol Ridge supports, which also rules it out for me.)
Yes, Oland/1st gen GCN sounds right. GCN 3rd gen is probably a typo or some error in the wikipedia page. https://en.wikipedia.org/wiki/AMD_Radeon_500_series#cite_note-16

I think a lot of these bottom of the barrel cards are used to fix ancient computers, where either the iGPU is outdated or the onboard video is broken and deteriorated from silicon wear or overheating. It probably is a shrinking market. When all the old stock of dies runs out I guess it will just get slightly more expensive to patch up old computers or add a second/third display.
 

ET

Senior member
Oct 12, 1999
513
26
91
I think a lot of these bottom of the barrel cards are used to fix ancient computers
These are purely mobile chips, thrown in by OEMs as slightly-better-than-iGPU solution, or, to put it another way, to trick customers into thinking there's a worthwhile discrete GPU in the package.

I'm pretty pissed with AMD for continuing to sell this piece of shit for so long. I know it's the easy and cheap way to go, but I want AMD to be better than this.
 

amd6502

Senior member
Apr 21, 2017
457
54
71
These are purely mobile chips, thrown in by OEMs as slightly-better-than-iGPU solution, or, to put it another way, to trick customers into thinking there's a worthwhile discrete GPU in the package.
That would be silly. (I don't doubt some would be unscrupulous enough to do this, but this decreases battery life and increases driver setup complexity).

But I totally forgot what maybe could be a significant and growing class of uses of cards lower than rx550 or 1030's:

People building HEDT workstation that want high CPU performance (eg Ryzen 1600x/2600 and above), but don't care about size of GPU. Like you mentioned in your prev post, I could see such people care about supporting modern standards and instructions.

If that market actually amounts to anything, then yes, why not introduce something modern, and save the leftover old GCN 1.0 for the very bottom end cards that use refurbed DDR3 (sodimm slotted memory).
 
Last edited:

dark zero

Platinum Member
Jun 2, 2015
2,507
8
61
Say Nosta... is possible to bring a Bulldozer version on Sub 5W chips to replace Mullins?
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,312
104
126
Say Nosta... is possible to bring a Bulldozer version on Sub 5W chips to replace Mullins?
On 28nm, no. On FDSOI, yes. Porting over Bulldozer/Excavator is more complex than porting across from Jaguar/Zen. With that port, Cluster-based multithreading can be completely mirrored. Mirrored means LSUs/Bus Units are equal distance to the Cache Unit/L2 for both cores. (No core with 2x the wire-length and no need for repeaters.)
----
https://i.imgur.com/9agyeN6.png
^-- my expectation on where XLP3 will be. The expectation is the nominal position(blue circle and the Volt-Freq line).

I also found more 22FDX/14FDSOI optimizations that are not in 22FDX, but are in 22FDX+. So, 22FDX+ is looking pretty high performance from 22FDX. So, confirmed perf is at the same performance of 12FDX(+26% perf and -47% power). So, 7LP perf for 22FDX+ which is cool, but there is also the unconfirmed perf/power enhancers as well.

I found a couple contractor semis that have taped out 22FDX IP potentially for AMD. There is also two new decode tests for AMD; AVS2 and AV1 for March'18. So, 22FDX+ APU and 12LP APU could possible support them in VCN 1.x?/2?.
 
Last edited:

ET

Senior member
Oct 12, 1999
513
26
91
That would be silly. (I don't doubt some would be unscrupulous enough to do this, but this decreases battery life and increases driver setup complexity).
Silly or not, that's what OEMs do. You get Bristol Ridge laptops with single channel RAM and Oland where dual channel would have about the same performance as the discrete chip (although Hybrid CrossFire does help increase performance beyond what either chip could do alone). I imagine that it helps OEMs ship a GPU with constant (even if low) performance regardless of the chosen configuration (APU choice and RAM configuration).

But from my experience it mainly serves to convince customers that there's a discrete GPU and therefore they will get something that's much better than an iGPU (most buyers' technical understanding extends to 'discrete is better' and 'higher numbers are better'). Which simply isn't the case.
 

dark zero

Platinum Member
Jun 2, 2015
2,507
8
61
Last edited:

NTMBK

Diamond Member
Nov 14, 2011
8,250
227
126
Unboxing Ataribox VCS: https://www.youtube.com/watch?v=pbYT3DyvLpc
A10 bristol ridge at ~30W, ram=4gb+free slot; bottom end zero frills starting price $200 ?
https://www.youtube.com/watch?v=AeSU63wQ-eU
https://www.youtube.com/watch?v=Q87L4WTBHSs
(Note: I've not watched the video yet due to being on my phone, so I'm just going off what you said.)

Entirely unsurprising. "Atari" clearly never had a big enough budget to pay for any new silicon. That said, it sounds like a nice budget APU gaming system. How much storage is installed in it?

I'm a bit sad they went with the VCS style case- I always thought that the Atari ST was the best bit of industrial design that Atari ever came up with. I'd love to see a modern AMD system in a case like that.
 

amd6502

Senior member
Apr 21, 2017
457
54
71
(Note: I've not watched the video yet due to being on my phone, so I'm just going off what you said.)

Entirely unsurprising. "Atari" clearly never had a big enough budget to pay for any new silicon. That said, it sounds like a nice budget APU gaming system. How much storage is installed in it?

I'm a bit sad they went with the VCS style case- I always thought that the Atari ST was the best bit of industrial design that Atari ever came up with. I'd love to see a modern AMD system in a case like that.
Well, if it catches on they could get parts revenue from selling alternate cases, and 2200g upgrade boards.

There are absolutely no hard specs out there, just hints of what the management is considering at this point (subject to change); such observations and sometimes speculation from bloggers. It does seem like it's more of a home computer video, music, and games streaming box. It could be very versatile and nice.

The unboxing vid is just humor. This group has a few other funny vids on their channel.
 
Thread starter Similar threads Forum Replies Date
moinmoin CPUs and Overclocking 29

Similar threads



ASK THE COMMUNITY

TRENDING THREADS