Trinity review

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
So, just to clarify, you're saying this because you know a lot about this industry?

Sort of like you said that people care more about CPU performance in mobile and yet people were and are still buying Llanos at higher rates than Intel alternatives, right?

I'm asking because your statements have a pretty solid history of being wrong. Do you have any legitimate proof that the rest of us can use?

LOL, keep deflecting the whole L3 cache argument. I'm awaiting your response.

And actually, 9/10 things I've said have been right. I was right about Kepler, I was right about Ivy Bridge (desktop), I was right about Sandy Bridge-E, and many other things.

BTW, just to inform you:

The max Turbo frequency for the HD 4000 will vary from models, but it'll be from 1100-1250MHz for standard voltage mobile. That's max, though, so it may not reach it if there's any thermal constraints. Clock scaling isn't that great for Intel's HD graphics, either. EUs/core configurations make a bigger difference, hence the difference I pointed out earlier of 5-10%.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
I'm not refuting anything with respect to the L3. The L3 is important and we both said that, I did with the gaming benchmarks and you did with Intel's massive lead in L3 read/write speeds. What you're dismissing, though, is that the HD4000, like the HD3000, relies very heavily on that cache, thus sharing it between CPU and GPU is likely to handicap it in certain scenarios -- for instance gaming, where they're both relying on the same L3 store. The L3 in Intel also works far more efficiently as far as reads/writes go, but Intel's approach copies the L2 into the L3 as it moves up a cache level. This is great for accessing it if you're speeds are synchronous (in comparison with AMD who has a very very slow asynch clocked L3), it also uses up more space to write the lower level L2 into the L3 store. AMD doesn't do this and instead relies on slower cache with essentially more "room" because it's L2 and L3 work independently. The core i5 IB's will have to rely on a 3MB cache that's split between both GPU and CPU as opposed to the 6(?)MB i7 that was reviewed has. So while it's 3MB theoretical room it's always less than that due to the L2 bumping up being stored in L3. What I'm saying is your "5%" figure is probably wrong, or you're pulling it out of you know what. I know what you're saying about cache, trust me.

What I don't understand is where you're coming up with any info about the i3. Thin air, maybe?
 
Last edited:

Olikan

Platinum Member
Sep 23, 2011
2,023
275
126
What's with the Skyrim "medium preset" difference between Techreport and Anandtech?

At Techreport the i7-3720QM HD4000 gets 24 Avg FPS at Anandtech it gets 42.7 (of what I assume is Avg FPS, they don't actually say)

Newer driver? HD4000 friendly benchmark run (Techreport runs around Whiterun one of the more GPU intense areas of the game)?

pcper...


??????
 

KompuKare

Golden Member
Jul 28, 2009
1,222
1,571
136
What's with the Skyrim "medium preset" difference between Techreport and Anandtech?

At Techreport the i7-3720QM HD4000 gets 24 Avg FPS at Anandtech it gets 42.7 (of what I assume is Avg FPS, they don't actually say)

Newer driver? HD4000 friendly benchmark run (Techreport runs around Whiterun one of the more GPU intense areas of the game)?

http://techreport.com/articles.x/22932/10

http://www.anandtech.com/show/5831/amd-trinity-review-a10-4600m-a-new-hope/6

Well Skyrim is hard to bench since there's no benchmark built in. I've benched it a few times but I use the intro and record it to log using MSI Afterburner. Seems the hardware sites do the whiterun 'run' up to Dragonsreach and look down at the square. For consistency it would be nice if some Skyrim mod writer made a comprehensive scripted bench mod and then got all the hardware sites to use it.

Another factor is of course the Skyrim version number. If one site is using a version prior to 1.4 that will make a lot of difference.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
Where are you getting this? I'd like to know. Intel hasn't released anything at all about the i3's other than a very very rough guesstimated release date of "Summer."
Thats because i3 may end up a high frequincy single core with 4000 igp and 3mb L3 cache. Iwould love to have such a chip in a lap top HT of course. say @25 watts
 

LOL_Wut_Axel

Diamond Member
Mar 26, 2011
4,310
8
81
Keep saying the mobile i3 won't have HD 4000 graphics and you probably won't admit a month from now that you were wrong.

Again, this goes back to what I said about Intel having the same product positioning hierarchy with Ivy Bridge as they did with Sandy Bridge.
 

pelov

Diamond Member
Dec 6, 2011
3,510
6
0
Keep saying the mobile i3 won't have HD 4000 graphics and you probably won't admit a month from now that you were wrong.

Again, this goes back to what I said about Intel having the same product positioning hierarchy with Ivy Bridge as they did with Sandy Bridge.

Correction, I'm not saying it won't, I'm just skeptical. Sort of how I'm skeptical whenever you say anything on these forums and assume it to be true. It's sort of like that
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
Pelov correct, lower Core i3 and Core i5 mobiles have less L3 than Core i7

I will also like to add,

SB and IB are sharing the L3 (LLC) between the CPU cores and iGPU via a Ring Bus. Also the L3 speed is connected to the CPU frequency. The higher the CPU core runs the faster the L3 becomes. It is like having a higher frequency DDR ram for the iGPU.

Even if Core i3 has the HD4000 (same number of EUs) L3 works in lower frequencies because Core i3 can turbo lower due to the lower TDP. That will have a huge impact in iGPU performance.

It is like having Llano with DDR-3 1333MHz and DDR-3 1866MHz.

So, Even if Core i3 or Core i5 have the same iGPU (HD4000) they will not have the same performance as a higher clocked Core i7.



 

sm625

Diamond Member
May 6, 2011
8,172
137
106
How did you arrive at your conclusion?

It is hard to find a compiled results since most review sites barely even review the i3, but you can see it here in this screenshot from anand bench:



There are many others like it. The sandy bridge quad, the full blown quad core 8 thread beast scores very well against llano, but the dual core i7s, not so much! I would expect the same from ib dual cores when they launch. Since the ib i3 wont even have turbo, I would expect an even larger difference between the i7 quad and the i3 dual.

The i7-37xxM is an extremely fast and powerful and expensive chip, and I continue to question the comparisons between it and AMD, and continue to point out the obvious lack of comparisons between the types of i3's and i5's you will actually find in notebooks priced near the AMD A10...
 
Last edited:

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
We have all ready seen the results of a 17watt intel and 2 core ivb@ 2.4ghz against trinity 4 core all 3 intel mobile chips were against the 4 core trinity the results speak for themselves. I gave the link . Trinity sucks compared to IVB. Many keep saying Intel drivers suck. Good I hope your correct . It there fore stands to reason Intel has the most to gain with driver updates.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
I think VR. zone did the best review . They cut threw all the BS and showed a review that places trinity perfectly. Against 2 core intels.
http://vr-zone.com/articles/amd-trinity-a10-4600m-apu-review-jumping-the-shark-/15830.html

From your link

System 1: AMD Trinity A10-4600 (35W Quad Core) with Radeon HD 7660G

AMD Comal Reference Platform

2 x 2 GB DDR3-1600 @ 11-11-12-28

CPUID Validation Link: http://valid.canardpc.com/show_oc.php?id=2363878



System 2: Intel Ivy Bridge (Dual Core) 2.4GHz with HD 4000

Core i7-3770K with half the cores disabled and downclocked to 2.4GHz (to emulate the i7-3517U for ultrabooks)

ASUS P8Z77-V Deluxe

2 x 4 GB DDR3-1600 @ 9-9-9-24

CPUID Validation Link: http://valid.canardpc.com/show_oc.php?id=2368926

Read more: http://vr-zone.com/articles/amd-tri...w-jumping-the-shark-/15830.html#ixzz1v3NvBJDs

Not only they have used tighter Ram timings but Core i7 3770K has double (8MB) L3 Cache than i7-3517U (4MB) cache.
 

Nemesis 1

Lifer
Dec 30, 2006
11,366
2
0
Yes I know its a simulation but the 17 watt sb is in there along with a modile from 2010 and I find the ram settings just fine as per intel spec.. So its a good review. Your the only one who thinks the L3 is game changer . I think it all depends how you slice it.
 

Khato

Golden Member
Jul 15, 2001
1,251
321
136
Have a look at the link and see what an impact the L3 has on gaming.

I would claim not much. Why? Because there are far more differences between the i5-2520M and i7-2820QM than just their L3 cache size. One is the obvious dual core with hyperthreading versus actual quad core, while another is the differences in maintained turbo speed due to the 35W vs 45W TDP, which is also responsible for the i5 having a dual core turbo speed of 3GHz compared to the i7's 3.3GHz.

Again, I'm not saying that the L3 size has no impact whatsoever an IGP gaming, merely that it's in line with the affects it has upon discrete GPU gaming as well. The cache is simply too small to have a more marked effect on the IGP.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
Unfortunately there is very little in the way of actual data that shows what effect ib cache size has on its IGP performance. I can only hope that when the dual core ib reviews roll out, they will take a 3770k and disable half the cores and clock it the same as the i3. I am willing to bet that L3 cache size will have a larger than usual impact on the ib IGP.

The test could be done right now if someone has both an i5-3570k and an i7-3770k. Just disable HT and clock them the same and do some benchmarks. But that jsut solidifies my point: those chips have sold thousands yet afaik not one test has been done to show what good that extra 33% cache does for the 3770k, with other factors eliminated.
 
Last edited:

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
while another is the differences in maintained turbo speed due to the 35W vs 45W TDP, which is also responsible for the i5 having a dual core turbo speed of 3GHz compared to the i7's 3.3GHz.

When i say look the impact SB/IBs L3 cache have on iGPU performance i mean both L3 cache size and speed.

As you correctly noted, a higher Watt TDP will allow for higher turbo speeds and that effects the L3 Cache as well. Im expecting that 35W IB chips with HD4000 will have 10% to 20% lower iGPU performance in games than 45W CPUs.

That will make the 35W Trinity 30-50% faster than 35W Core i3/i5.
 

Riek

Senior member
Dec 16, 2008
409
15
76
pcper...


??????

If i recall correctly, on anandtech it is tested on a large map/open space making it probably alot more cpu intensive than the others.

(also with the exception of skyrim and some issue in batman, HD4000 gets wiped pretty hard)

The difference is bigger than 680 and the 7950...
 
Aug 11, 2008
10,451
642
126
lol.....your gtx260 uses 75Watts alone, more than double of trinity max...

amd should realesa a 60 watt trinity

I never said it was efficient. He keeps the laptop plugged in when gaming anyway. The point is, APUs either from Intel or AMD have a long way to go vs a discrete card, even in laptops, when they cant match a 3 year old mid level gaming laptop. Obvioulsy, intel had a very long way to go and made progress, but not enough. AMD was ahead, but did not make enough progress with trinity to make it a game changer vs Llano or now even HD4000.
And calm down, AMD fans. I am not saying trinity is bad. They made progress on the CPU, but I was hoping for more improvement in the IGP, something like up to HD5670 levels.
 

KompuKare

Golden Member
Jul 28, 2009
1,222
1,571
136
Hm, sorry pelov and AtenRa but I just looked at some benchmark indexes of SB laptops (I used http://www.computerbase.de/artikel/...-envy-14-spectre/11/#abschnitt_3dmark_vantage) and I barely saw any 3D difference between the various benchmarked:

Using the average of 3Dmark Vantage 3D Performance, I got the following:
Core i3-2310M (2.1GHz base, 35W)________1240
Core i5-2467M (1.6GHz base, 17W)________1232
Core i5-2520M (2.5GHz base, 35W)________1383
Core i7-2620M (2.7GHz base, 35W)________2428

Unsure about the i7 (there were only two samples: 3204 and 1651 which seem strange). So it looks that unlike Llano HD3000 scores pretty much the same in all mobile SBs. Now IB is unknown but is I don't see it being that different.

It's the age old thing: Intel has the money to have the best fabs and since the P4 they're also using their money to have the best research. And of course the one time Intel's dominance was threatened (2003-2006 with the P4) they used their dominance and money for all kinds of dirty tricks.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
Check the numbers again KompuKare, you can clearly see that the i3s and the lower clocked i5s perform 15% lower in 3dmark vantage gpu score compared with the i7s and even the higher clocked (non-ULV) i5s. And this is on a somewhat meaningless benchmark. You have to look at a composite of actual gaming results because the intel igp is obviously specifically tuned to do well in 3dmark and similar tests. All the reviews show even the HD3000 competing well with llano in 3dmark but we know that isnt really the truth.
 

sm625

Diamond Member
May 6, 2011
8,172
137
106
Here is yet one more thing which muddies the waters:



That intel chip is (supposed to be) a 35W TDP. Obviously it is pulling much more power than trinity, which is also "35W". This is sort of telling us that AMD should be a LOT more aggressive with their turbo or else they are shooting themselves in the foot.

I wonder where this chips performance would be if they let their turbo run TDP up as high as to match the intel chip's power consumption? A simple linear extrapolation yields a potential 28% more performance!
 
Last edited:

MrTeal

Diamond Member
Dec 7, 2003
3,906
2,658
136
Any testing on a laptop is immediately suspect anyway, as there is a large number of variables that can't easily be controlled unless you can get two copies of the same laptop with different processors within them. That is doubly true of power consumption benchmarks. The best way to run the test would be to grab a 3770k and a 3570k. Disable Hyperthreading and overclock both to the same speed with no boost clock. At that point, you can say the difference is solely 8MB of L3 vs 6MB L3.
 

Olikan

Platinum Member
Sep 23, 2011
2,023
275
126
I never said it was efficient. He keeps the laptop plugged in when gaming anyway. The point is, APUs either from Intel or AMD have a long way to go vs a discrete card, even in laptops, when they cant match a 3 year old mid level gaming laptop. Obvioulsy, intel had a very long way to go and made progress, but not enough. AMD was ahead, but did not make enough progress with trinity to make it a game changer vs Llano or now even HD4000.

first of all, the APUs are meant to be cheap and power-eficient (well, cheap for OEMs)

second, a triple channel, 60Watt, trinity would beat a gtx260M...but this chip will never exisist, because of power and cost....
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |