AMD A10-7890K APU Review

[Alfredo_AMD]

Hello, everyone is miss the Point, it talking about AMD A10-7890K and it build in GPU performance VS Similar CPU Like AMD and Intlel I7 and Intel I3 with built in GPU or in the AMD Athlon X4 760K case without a Built in GPU and using a AMD R7 250 external GPU and there Performance of which the AMD A10-7890K came out at 2ed with the GPU at 1040 Mhz and 3rd with GPU at 866 Mhz both sharing DDR 3 Ram of the CPU just like Intel I7 and I3 CPU did and the AMD Athlon X4 760K with a AMD R7 250 GPU with 1 GB of DDR 5 Ram which came at 1st Place! As you can see here the AMD A 10-7890K APU is very good performance even beating Intel I7 6700K with built in GPU and Intel I3 with built in GPU which in this case not bad Considering Intel Cost more and I willing to beat that the AMD A10-7890K with a AMD R7 250 GPU and 1 GB of DDR5 Ram would get the same results as the AMD X4 760K and it AMD R7 250 GPU. Now if we where talking about CPU with a GPU Graphic Card like the Nvidia 750 TI we all know the Intel I7 6700K would be in 1st Place then AMD X4 760K 2ed followed by AMD A10-7890K in 3rd place each using a Nvidia 750 Ti GPU! Derf

 

[Central de Informatica]

AMD should:
1 Reduce the number of GPU cores.
2 Increase the GPU clock, offsetting the reduction.
3 Increase the CPU cache.

Thus it would reduce the bottleneck in memory access and gain CPU performance per clock cycle.

Hoping for better days AMD here in Brazil. Competition is always good.

Flavio Lima

 

[InvalidError]

AMD should:
1 Reduce the number of GPU cores.
2 Increase the GPU clock, offsetting the reduction.

If increasing GPU clock frequency was that easy, AMD and Nvidia would have done so already since higher frequencies are much cheaper to do than adding die area. GPUs are stuck around the 1GHz mark because that's as high as they are able to push the clocks before running into timing issues along critical paths.

 

[Central de Informatica]

AMD should:
1 Reduce the number of GPU cores.
2 Increase the GPU clock, offsetting the reduction.

If increasing GPU clock frequency was that easy, AMD and Nvidia would have done so already since higher frequencies are much cheaper to do than adding die area. GPUs are stuck around the 1GHz mark because that's as high as they are able to push the clocks before running into timing issues along critical paths.

It should not be so hard up about 20 to 25%. I live in Brazil, crazy heat, and with the internal GPU overclock my A10 with ease. Detail is that my PC is in the hottest spot of my house.

I with about 40% with stability. No increase in voltage. Fanless gamer or special case.

 

[InvalidError]

It should not be so hard up about 20 to 25%. I live in Brazil, crazy heat, and with the internal GPU overclock my A10 with ease. Detail is that my PC is in the hottest spot of my house.

I with about 40% with stability. No increase in voltage. Fanless gamer or special case.

Before AMD can do such a thing globally, they would have to ensure it works for 100% of people 100% of the time under the worst case conditions the chip is rated for throughout the warranty period. If AMD could already do that, they would already be making their chips that way to extract a few extra bucks from every sale and save a few bucks on every chip.

 

[porkysatan]

Go in BIOS disable support for legacy USB devices. Run the tests again. You will get much better results

 

[Central de Informatica]

It should not be so hard up about 20 to 25%. I live in Brazil, crazy heat, and with the internal GPU overclock my A10 with ease. Detail is that my PC is in the hottest spot of my house.

I with about 40% with stability. No increase in voltage. Fanless gamer or special case.

Before AMD can do such a thing globally, they would have to ensure it works for 100% of people 100% of the time under the worst case conditions the chip is rated for throughout the warranty period. If AMD could already do that, they would already be making their chips that way to extract a few extra bucks from every sale and save a few bucks on every chip.

AMD's central problem is in the 28 nm manufacturing process, already outdated. As she can not download the 28 nm trough of its responsible partner for the manufacture of chips she should try other partnerships pelop least to the top of the line chips of each follow-up.

It could achieve this with Samsung and other companies, but at least Samsung is already very involved with Intel, it sells notebooks.

But as I could not have to move. I think my alternative is valid. Even with little loss of GPU performance for the major bottleneck AMD is processing.

Another example: in my A4-4000 (32 nm / CPU 720MHz / 128 core) GPU ran at 2.0GHz. Without the temperature exceeding 53 ° C.

Then the bottleneck to increase the GPU clock can be thermal or electric power, but not physical.

It shows that a GPU more "simple" is easier to achieve higher clock rates.

 

[dreamalittle]

nice article, love the in depth benchmarks as per usual

MOD EDIT: Bad post deleted as requested. For future reference, you can just edit your posts.

 

[f688xt6]

It's pointless from a cost standpoint, but I'd like to see the comparison with a desktop Broadwell CPU like the i5-5675c to see if the APU holds up against Iris Pro at all.

 

[FormatC]

Take a look at my gaming benchmarks here, for example Need for Speed 2015 (a few days ago).
The Iris Pro 6200 is much faster - if the drivers are working. Intels software is mostly horrible.