I have a quick question, yeah I know I should of looked but...
Do the apu's even have hyper transport anymore?
What do the 890fx and 990fx use in wattage?
I thought it was around 45w's.
I have a quick question, yeah I know I should of looked but...
Do the apu's even have hyper transport anymore?
What do the 890fx and 990fx use in wattage?
I thought it was around 45w's.
Quote:
Originally Posted by justapost
nice spot.
I guess the result aligned with rumor that saying Kaveri has better INT performance and lower FP performance.
Thank you haylui,Quote:
Originally Posted by haylui
Results look good over all, for single core results there is an 20% increase in INT and FP performance. Odd thing with multi core the incerase is barely 10%. May be caused by the low memory speed or not yet finished tweaking of the core itself. I'm curious about the power consumption at the same clock in comparison to richland. Guess it can be abit higher on kaveri if their top model comes with an 3.7GHz baseclock at 95W, but that lower clockspeed may as well be an tradeoff to higher GPu clocks.
I was looking at the results of the Kaveri ES in geekbench and was wondering how it fares against Vishera.
This is my setup at the moment:
http://i.imgur.com/HflBInJ.png?1
and these are the results:
http://browser.primatelabs.com/geekb...aseline=223722
I was mainly interested in single core performance, and in my opinion it does pretty good. Multicore scores are lower, naturally.
Memory score seems too low, on the other hand (any thoughts on this?). My guess is that the chip was tested with basic 1600 mhz ram.
I will retest with my 8350 running at 3.5 ghz and ram at 1600 to see how it fares.
Followup
http://i.imgur.com/mt1r91A.png?1
results:
http://browser.primatelabs.com/geekb...aseline=243160
On single thread it's a win. On multithread, if we remove AES and Dijkstra out of the picture, the Steamroller CU's fare close or better to the Piledriver CU's (keep in mind the lack of L3 cache).
From the memory point of view i'm not sure what to make of it.
I think if we overclock these to 4.6-4.8 it should result in a faster chip than current visheras at least in single-threaded loads.
Not bad, I say.
Thanks for your test. But I feel sorry that your test makes no sense, because the Kaveri platform is running with single channel memory, and pulling down the result too much. I think you could make more test by running your FX8350 with single channel memory, and activating 4 cores only.Quote:
Originally Posted by chinook
Right, thanks for the heads-up, undone.
I've managed to disable 4 cores (this bios is not friendly when it comes to that) and swapped out one memory module.
http://i.imgur.com/TYItuW8.png?1
Results at 3.5 ghz 1600 mhz single channel
http://browser.primatelabs.com/geekb...aseline=243265
3.7 ghz
http://browser.primatelabs.com/geekb...aseline=243278
3.9 ghz
http://browser.primatelabs.com/geekb...aseline=243287
I could test at higher speed. Maybe kaveri benefits from ram speed more than vishera.
Actually it was only around 20 watts even maybe less now since that was around 790FX time.Quote:
Originally Posted by NEOAethyr
There have been power control upgrades to it since then.
no L3 cache effect this?
we need Richland A10-6800k vs this Amd Bantry system
Found a nice video of Kaveri BF4 demo:
http://www.youtube.com/watch?v=axyHkKn_e80
1920x1080 @medium 30-40fps vsync off (known numbers up to now)
1600x900 @medium 38-52fps vsync off
1280x720 @medium 55-60fps vsync on
1280x720 @high 30-35fps vsync off
1280x720 @ultra 25-30fps vsync off
That gpu should easily get a 40% OC (720 --> 1Ghz), do the math.
Its possible to reach 1600x900 high @30-35fps with some tweaks+gpu/mem OC
aaaah yaaah uuultra pleaase try...lol
i feel this igpu could definitely hit 1ghz OC clock. slap a decent cooler on it with some 2400mhz+ memory.
should see some demo leaks pretty soon! ;)
A couple of slides are now floating around showing 3.7 Ghz clock and 4.0 Ghz Turbo (max), so at stock settings it's looking like it will only be about 5% faster in single threaded applications (give or take depending on the nature of the code). It also is looking like there will only be a 15% increase in framerate (apart from mantle-enabled games). I'm sure that we'll see the following revisions with better clock speed, but the switch to bulk from SOI isn't treating them well this time around. The last thing to see is what sort of power consumption the newer cores have. We may not see a huge drop, but there may be some decrease (especially under load; idle is already quite good).
I don't believe that, for a few reasons.Quote:
Originally Posted by xVeinx
GF has never mentioned retooling for 28nm (or I just didn't read it)
TSMC has already done APU's with kabini and Tamesh on 28nm
There may be other explanations, but the point still stands that they aren't hitting the clocks they needed/wanted, and they did with Kabini and Temash. This shouldn't be new territory, so something went wrong, and it may involve the switch. Other suggested explanations are always welcome :).
the only intended clock speed I recall seeing was for FX being 4.0ghz 4.5ghz for Zamebezi chips.Quote:
Originally Posted by xVeinx
Since I don't think anyone answered you yet, no they do not. Hyper Transport is in the AM3 processors because they are the exact same die as the C32 and G34 Opterons. The hyper transport link is used to connect to the northbridge in the AM3+ boards. Server boards use the links to connect to NB and other CPUs. The APU's do not have a Hyper Transport because they are not meant to ever communicate with other CPUs in another socket and they have built in PCI-E lanes so those are used instead to connect to other chipsets.Quote:
Originally Posted by NEOAethyr
Can anybody tell me what the Opteron "Warsaw" is? http://www.xbitlabs.com/news/cpu/dis...rocessors.html
Could that be a new revision of Vishera?
tweaked piledriverQuote:
Originally Posted by Vassili
Posted before??
A10-7850K 4.9ghz 547 CB score
http://wccftech.com/amd-kaveri-a1078...mark-surfaced/
Real or fake?
What I learned so far is not to believe everything I see on the internetz :)Quote:
Originally Posted by thematrixhazune
to be fair, we don't know if NB or memory is clock at all or not.Quote:
Originally Posted by informal
it's about 20% slower than an I5 2500K vs the scores I've seen for 4.5-4.6ghz on cinebench 15.
Considering all AMD moves latelyQuote:
Originally Posted by Vassili
a) We may never see warsaw
b) it probably isnt a new revision, that would require new masks etc, if they were doing that they would probably do steamroller instead.More likely same core`s ,binned ,different setup.
Anyhow, wasnt there supposed to be some kind of kaveri preview this month ?
More bad news.
AMD roadmap for 2015 server wise:
http://www.youtube.com/watch?feature...&v=4Tf0wStiSXc
http://img801.imageshack.us/img801/4095/lq0k.png
So, after SR 4 cores there are 4 cored excavators with GPU.
What this essentially means, AMD doesnt want to build any non igpu more than 4 cores dies.If there would be anything bigger planned it would be in servers.AMD always doubled usage of chips from server to dekstop using same masks.
What is more troubling ,is that even jaguar cores vanish in exchange for ARM ones.
It essentially means that from NOW to END of 2015 there will be no bigger CPU from AMD than 4 cores.God dammit.
funny that it's called freedom fabric yet it's bga lol. (seattle cpu)
Hey, Toronto APU becomes a SoC! So Kaveri replacement will be new socket again, for DDR4 and SoC?
Also, thanks for that youtube vid vario, very interesting
I think Yes and No.
Look at information: DDR3/4
This could be as in Phenom II times (Phenoms can DDR2 and DDR3 too). So Excavator in desktop could working with FM2+ and FM3 boards. FM2+ are DDR3 and FM3 could be DDR4 boards?
Is it really necessary to change the socket? As long as they add the necessary traces on the motherboard, the processor, etc. ought to be able to handle DDR4...
I'm not so sure. If you listen to the talk he mentions that there will be improvements to the 6300 Opterons. One thing that was never fixed with the PileDriver core was the cache latency. That would not require a mask change. If they could deliver a "Piledriver2" with a modest 10% IPC increase in single thread and a big reduction in power usage (i.e. all desktop models at 95 watt) I would be satisfied enough to upgrade.Quote:
Originally Posted by vario
If i recall, Piledriver-->Trinity-->Richland. So perhaps fixing the cache and using Richland cores would give them 10% IPC?
Perhaps some ot the more knowledge folks in cpu desingn could enlighten us as to how much could be done without a big mask change.
Even a revision change requires new masks and process validation without any change to architecture.Quote:
Originally Posted by trueblue
A10% IPC gain is almost SR territory clock for clock.
"Improvements" and new SKU`s doesnt require any new silicon, just better/different binning,diff voltages/freqs.Great example are the "new" FX 9xxx .
If you look at the roadmap, keeping same architecture for 3+years means nothing new is coming,its essentially dead in the water just like am3+, there are going to be 2(!) major rehauls on the 1P side, SR and EX, while nothing on older G34 C32 server sockets.
Dont get me wrong new PD revision IS possible, it wouldnt bring any meaningful IPC gains, but it could yield higher freq or get power inline.But if they would do such a thing why not a SR/EX cored opterons from teh start.
New PD revisions will most likely come in the form of new steppings, I really doubt we will see any performance increase, only increases in power usage and maybe decreasing or the same overclock ability...trying to keep current with intel 22nm designs is a tough task for AMD :(
When we look at how AMD "left the high-performance market" we can see that it starts with server side...
Steamroller will not come to a high core design, the 6300 series is supposed to last through 2015?
AMD will lose a ton of market share in that market to intel, and already are...while the 12 and 16 Core Opterons (6100-6200 series) made sense in 2011, they make little to no sense at all in 2013, with Ivy Bridge-EP and upcoming Haswell Xeon (up to 18 core, 36 thread) designs...
Intel is putting AMD to shame and AMD is surviving on sales of products in the $50-150 range which have slim profit margins...
Anyone remember Terramar? Up to 20 cores? Cancelled? No PCI-E 3.0? :rofl:
Will Kaveri support quad-channel RAM?!
When looking at the QVL for RAM for this gigabyte FM2+ board, 4-channels is mentioned: http://download.gigabyte.eu/FileList...niper-a88x.pdf
When looking at an older FM2 gigabyte board, this same QVL list only goes up to dual-channel: http://download.gigabyte.eu/FileList...f2a85x-up4.pdf
What do you guys think?
I think that's some sort of an error in that document. Maybe they mixed up data meant for some intel s2011 motherboard.
http://www.chiphell.com/forum.php?mo...e=1&mobile=yes
Kaveri scores in cb r15...
See first thread for scores
links dead.Quote:
Originally Posted by haylui
the score is too low anyways.
Fixed.Quote:
Originally Posted by demonkevy666
Btw could this be real score?
My guess is no, but I would love to be surprised.Quote:
Originally Posted by Musho
Link is still dead... Or they killed the thread.Quote:
Originally Posted by haylui
Quote:
Originally Posted by informal
See the link below
The first post..
http://www.chiphell.com/forum.php?mo...&extra=&page=1
Ah ok I see it now. Well this looks more plausible than that BS number of 547pts @ 4.9Ghz ;).
Original link is :
http://www.gdm.or.jp/voices/2013/1230/56219
no it's not, that only it put it's cpu 3-7% faster than richlandQuote:
Originally Posted by informal
So let me get this straight: if you personally don't like the result it must not be true?Quote:
Originally Posted by demonkevy666
We need to get back to reality.
just where do you get that from? you said it was plausible. I'm saying it's not plausible. the said cpu increase over richland was 20% on cpu side.Quote:
Originally Posted by informal
1st of all that slide was marketing slide.Quote:
Originally Posted by demonkevy666
2nd it was about mobile Kaveri.
3rd it stated "projections not based on actual silicon". So it was based on simulations or AMD wasn't sure where the clock will land so they left that door open.
4th they listed zero workloads in which it was supposedly going to be "20% faster" (based on simulations!).
To summarize, that mobile related slide with "20% faster on CPU side" bullet point is basically telling us nothing about how actual desktop Kaveri will perform.
disabling a core in piledriver module already shows an 18-22% increase in performance, with out the decoder bottleneck.Quote:
Originally Posted by informal
Disabling a core in module is completely different than "adding another decoder". IN 1st case you have whole front end working for that one core and a whole FP unit (2x FMAC) working for that one core.Quote:
Originally Posted by demonkevy666
In case of SR, you have still a shared fetch unit, increased but still shared Instr. cache and the same or even less FP resources per "core" than in the case of disabled core in a PD module. As can be seen, adding a decoder in no way makes a SR "core" on the same level as PD module that has one core disabled.
lataer today I can try Cinebench R15 with 1600MHz DRAM A10-6800K (as this Kaveri) for the same score. I think, it will be around 4200-4300 MHz CPU clock.
I already have tried it, ~4.2Ghz PD scores about the same 311pts :). At fixed 3.7Ghz clock 4T PD (Richland) scores 274pts.
OK, there is it. First one is A10-6800K with 2133 MHz DDR3 and windows set to performance. CPU cooler Noctua NHD14, base clocks at stock. In Cinebench I was looking for frequencies - these was between 4200-4300 MHz at cores (A10-6800K is 4.1/4.4 GHz chip with better power state modes than Trinity). 324 points.
http://i.imgur.com/fDthluu.png
But if disable power control, I tried 4100 MHz. And lucky 318 points :). So the same as Kaveri at stock settings. The memory at Richland are the same 1600 MHz and I suppose cl9 timings.
http://i.imgur.com/bXEjkm7.png
So, we can say simply the performance (Cinebench) of stock Kaveri 3.7/4.0 GHz is the same as stock Richland A10 4.1/4.4 GHz. IPC is better around +300-350 MHz.
That's less than 10% improvement :(...
10% is good. Look at Intel. SB vs IB around 3%, Haswell vs IB around max 5%. Is not much possible with x86 get higher. Only in software with new instruction support level (FMA, AVX)
Understood, I was simply hoping that the performance ceiling (with changes) for the architecture would be closer to Haswell, and that they would be able to make some larger gains up front with SR to get performance at least to the level of Nehalem.
Thanks for the comparison FlanK3r, very interesting. Could you run the test in sigle core at same clockspeed to make a direct comparison between the architectures? In multithreaded benchmarks the 7850K should have the advantage because of the changes in the scheduler.Quote:
Originally Posted by FlanK3r
Thanks again!
I did a quick calculation based on FlanK3r's results. Perf gain is 8.37%. To be honest, from all the leaks that has appeared before for AMD, I think we can trust these results.
((311/3.7) * 4.1 ) / 318
311 is multithreaded CB for Kaveri, 3.7 is Kaveri's clock speed, 4.1 is Richland clock speed and 318 is multithreaded CB for Richland.
First is 4100 MHz 6800K and second screenshot is stock A10-6800K with turbo.
http://i.imgur.com/4bksC0x.jpg
http://i.imgur.com/l1QYEL8.jpg
news about kaveri on planet3dnow.de
A full spread of benchmarks are collected (originally from Baidu) by WCCFtech: benchmarks
There are some wild differences at points (6 second drop in Super Pi?), while gaming sees little improvement over 6800k.
crazy is wprime too. But I tried BD fix for Richland and superpi is
http://i.imgur.com/cwrtDws.png
So 3s difference against Kaveri
Seems as if there is maybe not much to look forward to then?
I think we'll need to see more real world benchmarks to know either way, but it looks like some nice improvement for a select few apps. Otherwise, its HSA/Mantle/TrueAudio that headline for this APU. More slides here: slides.
Oh, and pricing is at $173 now...
WccFtech forgot to mention like the original post has stated that the A10-7800 is just ENGINEERING sample
Flank....need your advice, 1st of ALL HAPPY NEW YEAR...Quote:
Originally Posted by FlanK3r
I always remembered you talked about the Gskill TridentX is Super good with AMD.......just ordered this :
http://www.newegg.com/Product/Produc...ab=true&Page=2
I wish they had 2x4 8gb but with the Cas7 in 1600mhz only have 2x8gb/
will 16Gb of ram 2 sticks slow me down vs 8gb?
thanks Flank
less populated slots = less latency and less stress on imc and thus better for overclocking.
very good news for me thanks, right now I have 4 slots at 2gb each 8 gb total......so now I will have 2 slots at cas7 vs my OCZ cas8 4 sticks......Quote:
Originally Posted by i found nemo
needed a little kick of encouragement...thanks nemo
Found these over at Hard
http://imgur.com/a/XVob4
Quote:
Originally Posted by Darakian
thanks good find......:up:
for all testers, what need CPU-Z support for Kaveri. Fresh beta: http://www.cpuid.com/medias/files/so.../cpuz_1673.zip
Quote:
Originally Posted by FlanK3r
thank you flank
:up:
just got these tridentX gskills cas7 working good -2x8gigs very very fast and tight.......very impressive!
Nb is over clock on second picture it's also worth about 5% increase in single thread on cinebench 15.Quote:
Originally Posted by FlanK3r
what's the highest over clock on the nb that's stable ?
no, its P-state. UNB clock in idle going down from 1800 MHz
oh I see it's turbo is on too.Quote:
Originally Posted by FlanK3r
Hello! Where can i find drivers for my 7850k? Amd auto detect doenst work and there is no drivers anywhere :(
I cant use it igp. This is my personal review:
http://foro.noticias3d.com/vbulletin...d.php?t=421821
download drivers from motherboard page (in chipset), its working.
It works! Thanks!
P.D is a heavy compute performance than trinity. Double in luxmark, tripple in sph fluid simulatiom and raytrace.
Thank you for the preview djohny. Indeed ocl results are great. But these require Windows 8 atleast for full HSA support? Also thank you flank3r for your comments at overclock.
Thank you for that test!Quote:
Originally Posted by djohny
I can't wait to upgrade my A10-5800K to Kaveri! At least now I can see what a massive jump it will be, especially in compute and gaming on discrete graphics card :)
Quote:
Originally Posted by djohny
you mean Kaveri?
http://imgsrc.baidu.com/forum/w%3D58...4e251f5850.jpg
http://imgsrc.baidu.com/forum/w%3D58...1f95ca5f50.jpg
http://imgsrc.baidu.com/forum/w%3D58...cad1c85e50.jpg
The 28nm process looks just as fast as the 32nm SOI looking at anandtech's numbers. It's only 1 sample, but it looks like the clock speed drop due to increased power draw due to now having transistor body leakage since the SOI layer is gone. Too bad SOI got dropped for 28nm. If it was still there I'd expect the CPU clocks to be the same as richland, if not slightly higher.
Overall I'm disappointed with kaveri. The place where the APU makes the most sense is to save costs and board space on laptops while exceeding the GPU performance that intel can offer without a dGPU. While the CPU performance on desktops is OK for games, on laptops it will be lacking. Kaveri is going to run 3ghz at best in a laptop, which is going to have issues maintaining vysnc in games (my 4600m does when tweaked so it performs better than a 5750m). Yeah you can max out the resolution and run at 30fps instead, but to be forced to do that is ridiculous. GPU performance does look good though when it isn't being crippled by the CPU.
Also the gains in 512 shaders in the 7850k over the 384 in the 7600 are often less than 10% despite a 50% stronger core. When not CPU limited kaveri is badly memory starved.
http://www.anandtech.com/show/7677/a...0-a10-7850k/12
It's a bit offtopic and all, but there is a petition now by Simeon Toshev:https://www.change.org/petitions/adv...itecture#shareQuote:
To:
Advanced Micro Devices (AMD), One AMD Place P.O. Box 3453 Sunnyvale, CA 94088-3453
Please release high end FX processor with steamroller microarchitecture. Steamroller is already in the new APUs called "Kaveri" so it's a big shame not having it in a high end FX CPU because a lot of people are waiting for Steamroller FX high end CPU.
I hope you consider the release of high end Steamroller FX CPU as it'll make many people very happy!
Sincerely,
[Your name]
I thought you people might like it
if people are disappointed with kaveri They'd be even more disappoint with steamroller FX (see below)Quote:
Originally Posted by donitsi
Quote:
Originally Posted by Khenglish
increase in power is coming from going from two (4 way) decoders in richland to four (4 way) decoders in kaveri/Steamroller.
so increasing to eight (4way) decoder is not a good idea.
what we need is rework of how the module uses single thread
signed, though i'm doubting it'll do well. just hope i'm wrongQuote:
Originally Posted by donitsi
Say Amd starting tomorrow decided to do a limited run. Wouldn't it take like at least 6 months to a year to get it going?
Id rather have them make a 6-8 core apu with a smaller igpu