AMD: Sea Islands R1100 (8*** series) Speculation/ Rumour Thread

As you may have heard, AMD has officially released Richland, the APU that replaces Trinity. It improves on Trinity's Turbo by taking temperature under account to take advantage of any available headroom, like Intel's Turbo or NVIDIA's boost. More at the Tech Report: http://techreport.com/news/24482/amd-intros-35w-richland-mobile-apus

The reason I'm posting this here is that if AMD is doing this for APUs, I don't see why they wouldn't do the same for GPUs at some point either this year or in 2014. The obvious downside is that the new algorithm isn't deterministic anymore, but I guess AMD figured that an extra 2~10% or so in benchmarks was worth it.

I believe AMD have started introducing a boost feature that does something like this, it is available on the 7900 series chips.
 
Reviewers should start doing hot box tests, like some power supply reviewers do. Run the system in a poorly ventilated case of yesteryear in a non-ac room and compare the results. It's a shame that people can't be certain that they will get the performance they see in reviews from a product they buy, but I guess it's better now that all the big players will be on the same playing field...

I think this is one of the main reasons that Intel Core series CPUs are so conservatively clocked. By clocking them so conservatively it basically makes them deterministic. As even in the worse conditions at stock settings you should still get the same CPU boost as anyone else.

Regards,
SB
 
I think this is one of the main reasons that Intel Core series CPUs are so conservatively clocked. By clocking them so conservatively it basically makes them deterministic. As even in the worse conditions at stock settings you should still get the same CPU boost as anyone else.

Regards,
SB

It also makes their cpu's look better for a given tdp. That's fine though as long as the end user can expect to actually get close to this performance in their laptop on a hot day. It's when we have things like Intels new SDP that it gets grey, as the delta between min and max clock speeds is massive.
 
As you may have heard, AMD has officially released Richland, the APU that replaces Trinity. It improves on Trinity's Turbo by taking temperature under account to take advantage of any available headroom, like Intel's Turbo or NVIDIA's boost. More at the Tech Report: http://techreport.com/news/24482/amd-intros-35w-richland-mobile-apus

The reason I'm posting this here is that if AMD is doing this for APUs, I don't see why they wouldn't do the same for GPUs at some point either this year or in 2014. The obvious downside is that the new algorithm isn't deterministic anymore, but I guess AMD figured that an extra 2~10% or so in benchmarks was worth it.

I was a bit disappointed. As soon as I saw the announcement, I came here to see what the Beyond3Ders had to say about it, but even after doing a forum search I didn't see much mention of it. Is it really that un-exciting, or is there not enough info out yet to discuss?
 
It's a tweak of Trinity with some firmware updates. Some of the most significant of its changes, such as the thermally-influenced turbo, rely on hardware that was already in Trinity, but not fully utilized. Thermally-influenced turbo has been around for multiple CPU generations by somebody else who makes better CPUs, so it's not new.
Someone at AMD had the sense of humor to have a marketing picture of Trinity next to Richland, if they didn't just copy-paste the same picture.

There have been more noticeable changes within the same product line, such as in the Athlon XP days with the Thoroughbred A and B revisions.
If AMD were working on its original schedule, Richland wouldn't exist or would have been Trinity in the first place and we'd be looking at Steamroller right now and looking for rumors on Excavator.

I guess we could speculate as to how many non-crummy laptops Richland will wind up in before Haswell debuts.
 
What 3dilettante said, plus almost everything that AMD revealed about Richland was already known, with the exception of the temperature-based Turbo.

I do think the 25W and 17W versions of this APU will be quite nice, though.
 
Yeah there's not much to discuss. Some years ago they'd have called this a new stepping instead of a new platform...
I agree with Alexko that it might make for potentially nice lower power chips. But to discuss that actual hw review would be required.
 
If AMD were working on its original schedule, Richland wouldn't exist or would have been Trinity in the first place and we'd be looking at Steamroller right now and looking for rumors on Excavator.
Word on the streets is the original Kaveri APU had a 'Steamroller A' core. This product was canceled in favor of a 'Steamroller B' core based one which had been planned as a Kaveri's refresh.

Did they want the perf. delta between PD and SR based products to be bigger? Was SR A a failure? Was SR B project ahead of schedule? Who knows...

Anyway, they have created a new revision of Trinity as a stop gap product - their marketing dept. calls it Richland :)
 
What 3dilettante said, plus almost everything that AMD revealed about Richland was already known, with the exception of the temperature-based Turbo.

I don't think so:

Footnotes:

2. Testing and projections develops by AMD Performance Labs. The score for the 2012 AMD A10-4600M on the Cinebench Multi CPU benchmark was 2.0 and the 2012 AMD A8-4555M was 1.30, while the "Richland" 2013 AMD A10-5750M was 2.35 and the 2013 AMD A8-5545M was 1.55. PC configurations based on the "Pumori" reference design with the 2012 AMD A10-4600M with Radeon™ HD 7660G graphics, the 2012 AMD A8-4555M with AMD Radeon™ HD 7600G graphics, the 2013 AMD A10-5750M with AMD Radeon™ HD 8650G graphics and the 2013 AMD A8-5545M with AMD Radeon™ 8510G Graphics. All configurations used 4G DDR3-1600 (Dual Channel) Memory and Windows 7 Home Premium 64-bit. RIN-3

According to AMD the A10-5750M is ~17.5% faster than the A10-4600M in cinebench 11.5

At least for me that was a nice surprise. Could be related to the Turbo or further bug-fixing.

Link: http://www.planet3dnow.de/vbulletin/showthread.php?t=411513
 
I don't think so:



According to AMD the A10-5750M is ~17.5% faster than the A10-4600M in cinebench 11.5

At least for me that was a nice surprise. Could be related to the Turbo or further bug-fixing.

Link: http://www.planet3dnow.de/vbulletin/showthread.php?t=411513

Interesting that IPC seems improved: looks like the clock speed boost is under10% (2.3 / 3.2 to 2.5 / 3.5 ghz) where as the boost in performance is quite a bit higher; wonder what's been tweaked here.

It could also be the case that the CPU is spending more time near 3.5 ghz because of better power-tune than the older CPU is spending at 3.2 ghz.
 
Interesting that IPC seems improved: looks like the clock speed boost is under10% (2.3 / 3.2 to 2.5 / 3.5 ghz) where as the boost in performance is quite a bit higher; wonder what's been tweaked here.

It could also be the case that the CPU is spending more time near 3.5 ghz because of better power-tune than the older CPU is spending at 3.2 ghz.

I think the latter option is correct. Some minor bug-fixes could result in marginal IPC improvement, but it's probably mostly the new Turbo.

Besides, if it really is primarily temperature-driven, then TDP is more of an afterthought.
 
Yeah there's not much to discuss. Some years ago they'd have called this a new stepping instead of a new platform...
I agree with Alexko that it might make for potentially nice lower power chips. But to discuss that actual hw review would be required.

Ah, now I understand. :) Now I know why there is less discussion of steppings now a days... >=P They sometimes name them as new chips!
 
Seems like a ghost town in here.

What are the chances that the new 7790 card gets rebadged as a 8750 or 8770 card?

I hope that it doesn't get rebadged as a 8770 as then we will have 7770/7850 performance and bandwidth gap again.

8750 should be an underclocked or cut down 7790(for lower TDP) with the same 1GB GDDR5 @ 6Gbps for 96GB/s bandwidth.
8770 Should be a 192bit bus with 1.5GB GDDR5 @ 6Gbps for 144GB/s bandwidth.
8850 should be 256 bit bus with 2GB GDDR5 @ 6Gbps for 192GB/s bandwidth.

If the 8770 has 128 bit bus the bandwidth will be half the 8850, which is way too large of a gap, that's why i think 192 bit makes sense.

Thoughts?
 
They needed 2.5 years to put the 6870 type of performance to the 7790 level. :LOL:

I hope 7790 is equal to 8570, 8630 or something.... It's very sad to me to see such low performance improvements.
 
They needed 2.5 years to put the 6870 type of performance to the 7790 level. :LOL:

I hope 7790 is equal to 8570, 8630 or something.... It's very sad to me to see such low performance improvements.

You can't be serious.

The 7790 uses about 50% less power and costs $90 less at launch. The reason for that being simple enough - it's a tier below the 6870.
 
Seems like a ghost town in here.

What are the chances that the new 7790 card gets rebadged as a 8750 or 8770 card?

I hope that it doesn't get rebadged as a 8770 as then we will have 7770/7850 performance and bandwidth gap again.
At least in the OEM space, there is a significant numerical gap in naming between the 8760 and the 8870, so I think the 7790 could find its way in there. I would even guess that this specific OEM numbering may have been chosen in part because they wanted a space for parts based on the upcoming Bonaire.
 
You can't be serious.

Oh, just sit comfortably in your armchair and watch.

The 7790 uses about 50% less power

This is good, of course, but it is so trivial that I can't explain it. Every die shrink on a new process achieves it. Also, this card is not aimed at people owning 6870 because the most important is the raw performance... And after that if someone is willing to count his cents in the wallet, how much he/ she will save via a better card.

costs $90 less at launch. The reason for that being simple enough - it's a tier below the 6870.

No, it costs so much less because with time this performance naturally loses value. ;)
 
Oh, just sit comfortably in your armchair and watch.


It's always happened. 7600GT has the same performance as 6800GT, gtx 560ti same as gtx 470, geforce 2MX same as geforce 1 sdr, voodoo3 2000 same as voodoo2 SLI.
These cards were very successful and well received, too.

When my parents bought a new PC over 16 years ago, was I pissed because it had a 6x86, S3 virge and IDE disk rather than older Pentium Pro 150, Matrox Millenium and SCSI? Hell no. It wasn't that worse, and to get the latter they would have had to sell a car or make us eat only pasta.

You shouldn't be using a PC at all. The PC originally came with the 8088, the original Celeron so to speak, and the crappy graphics was a costly option, you could get the graphics card that only does text instead.
 
Last edited by a moderator:
Back
Top