66AK2H14: The max frequency on 66AK2H14 and 66AK2E05

Hi Jacky 

For #1 Please look at the datasheet "device nomenclature" section ,  you can order parts that run at DSP/ARM at 1.2GHz or DSP/ARM at 1.2GHz/1.4GHz. The DSP is always 1.2GHz, for ARM we offer both 1.2 GHz and 1.4GHz variants. 

For #2 Please refer to the datasheet intro section and device nomenclature section. We offer ARM/DSP running at 1.25GHz or 1.4GHz. 

For #3  it is the superset CPU , 66AK2H14 and should allow maximum frequency evaluation. 

Hi, Mukul

Thanks for reply.

For #1 I check  "device nomenclature" section, 66AK2H14AAW mean 1GHz. It seems don't match the device masking(1.2GHZ/1.4GHZ)

I share the data sheet definition and my IC capture.

Do you have idea about this?

For #2 I did refer to  device nomenclature section, but I could get last number form device mask(don't see ABD25 or ABD4)

Is it special version of my K2E IC?

For #3 Could you share the part number of EVMK2HX CPU, please? Many thanks.

Best Regards

Jacky

Hi,Mukul

Mukul Bhatnagar said:

For #1 Please look at the datasheet "device nomenclature" section ,  you can order parts that run at DSP/ARM at 1.2GHz or DSP/ARM at 1.2GHz/1.4GHz. The DSP is always 1.2GHz, for ARM we offer both 1.2 GHz and 1.4GHz variants. 

How can I know the ARM core running on 1.2GHz or 1.4GHz? Does K2H have register to log the current  core frequency?

Or have other CPU pin to output system clock?

Best Regards

Jacky

Jacky

May I ask what is the problem/debug challenge you are facing? If you ordered 1.4GHz capable parts they should run at 1.4GHz. 

Please see if the following thread and the gel files/register description help

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/858629/evmk2h-enquiry-about-evmk2h-clock-frequency

Regards

Mukul 

Hi,Mukul

I did face some problem I could not figure out .My team use same program to test the performance on both K2H and K2E IC.

They find K2E is slower about 1.5s than K2H on every 28s.

I want to confirm why K2E is slower?

Do you have advise for me to check? If K2H is running 1.2GHz, what else will cause test time difference?

Best Regards

Jacky

Dear Customer,

Good day!

                    "Do you have advise for me to check? If K2H is running 1.2GHz, what else will cause test time difference?"

Yes. There is a method, by which we can get to know the device speed whether it runs at 1.2GHz or not. 

Please use the gel file below, load it in CCS and run it in your platform. This will output the informations such as DEVSPEED, TAGID, DEVSTAT, DIEID of your device.

6266.K2H_register_dump.gel

Can you please use the GEL in CCS  (In CCS Debug View  Tools->GEL files; Right click and import the GEL attached ). You can then go into Scripts-> K2H Device Efuse settings  and provide the log here.

Repeat the same procedure for both the boards, K2H and K2E, provided, the appropriate memory addresses are used as per the K2E and K2H data sheets.

------

OR

We can directly use the memory address (0x02620C98 - for K2H) of the DEVSPEED register in CCS memory browser window and get the output. For K2E, refer the data sheet and use the appropriate memory address for the DEVSPEED.

--------

OR

Use the SYSCLKOUT pin which gives the SYSCLK value/6. 

For 1.2 Ghz PLL settings, should read 200 Mhz on that pin.

Similarly, calculate for K2E.

Note: You can also find the same suggestions in this thread too with more explanations: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/858629/evmk2h-enquiry-about-evmk2h-clock-frequency

Regards

Shankari

Hi,Shankari

Thanks for detail reply and method you supply.

For #3

Shankari G said:

Use the SYSCLKOUT pin which gives the SYSCLK value/6

            I check  SYSCLKOUT  output

            I could see K2H is running on 1.2GHz, K2E is running 1.25GHz

For #1

Shankari G said:

Please use the gel file below, load it in CCS and run it in your platform. This will output the informations such as DEVSPEED, TAGID, DEVSTAT, DIEID of your device.

From the log I could see K2H and K2E both run 800MHz.

It is strange that it doesn't math the SYSCLKOUT pin output frequency.

I think maybe something wrong on my configure? Do you  any advise for me to check about that?

Here is the log on both K2H and K2E using your Gel file.

CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************
CS_DAP_DebugSS: GEL Output:  ********************************** K2E EFUSE STATUS and SNAPSHOT ******************************
CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************

CS_DAP_DebugSS: GEL Output:  ********************************** K2E DEVSTAT Register (DEVSTAT) ************************************

CS_DAP_DebugSS: GEL Output:  DEVSTAT ---> 0x00008005
CS_DAP_DebugSS: GEL Output:  ********************************** K2E JTAGID Register (JTAGID) ************************************

CS_DAP_DebugSS: GEL Output:  BOOTCFG_JTAGID ---> 0x0B9A602F
CS_DAP_DebugSS: GEL Output:  ********************************** K2E MACID Register (MACID) ************************************

CS_DAP_DebugSS: GEL Output:  MACID[31:0] ---> 0x4C0CDE07
CS_DAP_DebugSS: GEL Output:  MACID[32:47] ---> 0xB499
CS_DAP_DebugSS: GEL Output:  BCAST[16](Broadcast Reception) ---> Broadcast
CS_DAP_DebugSS: GEL Output:  BCAST[17](MAC Flow Control) ---> Off
CS_DAP_DebugSS: GEL Output:  CHECKSUM[24:31] ---> 0x19

CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************
CS_DAP_DebugSS: GEL Output:  ********************************** K2E DEVICE SPEED CONFIGURATION ******************************************************
CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************

CS_DAP_DebugSS: GEL Output:  BOOTCFG_EFUSE_RSVD2 ---> 0x00000000
CS_DAP_DebugSS: GEL Output:  DEVICE SPEED[27:16] ---> 800 MHz
CS_DAP_DebugSS: GEL Output:  ARM SPEED[27:16] ---> 800 MHz
CS_DAP_DebugSS: GEL Output:  ********************************** KS-II DIEID Register (DIEID) ************************************

CS_DAP_DebugSS: GEL Output:  DIEID0 ---> 0x05013013
CS_DAP_DebugSS: GEL Output:  DIEID1 ---> 0x0C00EA89
CS_DAP_DebugSS: GEL Output:  DIEID2 ---> 0x00000000
CS_DAP_DebugSS: GEL Output:  DIEID3 ---> 0x4A6E0000

CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************
CS_DAP_DebugSS: GEL Output:  ********************************** K2H EFUSE STATUS and SNAPSHOT ******************************
CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************

CS_DAP_DebugSS: GEL Output:  ********************************** K2H DEVSTAT Register (DEVSTAT) ************************************

CS_DAP_DebugSS: GEL Output:  DEVSTAT ---> 0x02012005
CS_DAP_DebugSS: GEL Output:  ********************************** K2H JTAGID Register (JTAGID) ************************************

CS_DAP_DebugSS: GEL Output:  BOOTCFG_JTAGID ---> 0x2B98102F
CS_DAP_DebugSS: GEL Output:  ********************************** K2H MACID Register (MACID) ************************************

CS_DAP_DebugSS: GEL Output:  MACID[31:0] ---> 0xEA5E730C
CS_DAP_DebugSS: GEL Output:  MACID[32:47] ---> 0x74DA
CS_DAP_DebugSS: GEL Output:  BCAST[16](Broadcast Reception) ---> Broadcast
CS_DAP_DebugSS: GEL Output:  BCAST[17](MAC Flow Control) ---> Off
CS_DAP_DebugSS: GEL Output:  CHECKSUM[24:31] ---> 0x29

CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************
CS_DAP_DebugSS: GEL Output:  ********************************** K2H DEVICE SPEED CONFIGURATION ******************************************************
CS_DAP_DebugSS: GEL Output:  *******************************************************************************************************

CS_DAP_DebugSS: GEL Output:  BOOTCFG_EFUSE_RSVD2 ---> 0x00000000
CS_DAP_DebugSS: GEL Output:  DEVICE SPEED[27:16] ---> 800 MHz
CS_DAP_DebugSS: GEL Output:  ARM SPEED[27:16] ---> 800 MHz
CS_DAP_DebugSS: GEL Output:  ********************************** KS-II DIEID Register (DIEID) ************************************

CS_DAP_DebugSS: GEL Output:  DIEID0 ---> 0x0900D00D
CS_DAP_DebugSS: GEL Output:  DIEID1 ---> 0x0C0000CD
CS_DAP_DebugSS: GEL Output:  DIEID2 ---> 0x08000000
CS_DAP_DebugSS: GEL Output:  DIEID3 ---> 0x17000000

For 2# 

I also check the DEVSPEED register(0x02620C98) on both K2H and K2E. 

The register did show K2H and K2E both run 800MHz.This match  your Gel file output.

So I confuse that why SYSCLKOUT pin don't match DEVSPEED register?

Best Regards

Jacky

Hello Jacky 

Can you share a few more details 

1) Are these TI EVMs or custom boards?

2) If it is custom board, can you share the details on the PO for the units you have purchased?

3) In an earlier post you had shared the snapshot of the case marking for K2H - can you confirm that the gel file is read out from the same unit?

Additionally please look at the DEVSPEED and ARMSPEED field definitions in the K2H datasheet and it would be good for you to reconfirm that value of these registers, instead of the gel file read out (I am not sure if the gel file is dated /old) 

Pg 226: https://www.ti.com/lit/ds/symlink/66ak2h14.pdf 

Hi,Mukul

Mukul Bhatnagar said:

1) Are these TI EVMs or custom boards?

For K2H , this is the TI  EVM board ( TI EVMK2H and EVMK2HX)

For K2E,  this is the custom boards. but base on EVMK2E, just cut some hardware, such as PCI to SATA controller.

Mukul Bhatnagar said:

2) If it is custom board, can you share the details on the PO for the units you have purchased?

For now I only could share the part number of K2E IC : 66AK2E05XABD25

Mukul Bhatnagar said:

3) In an earlier post you had shared the snapshot of the case marking for K2H - can you confirm that the gel file is read out from the same unit?

Yes, it the same K2H unit which is form  TI EVMK2H board.

Mukul Bhatnagar said:

Additionally please look at the DEVSPEED and ARMSPEED field definitions in the K2H datasheet

I check the register on both K2H and K2E.

K2E                                                                               K2H

Form register definition show , K2E and K2H did run at 800MHz, which still do not match SYSCLK pin output.

Do you have any advise?Thanks

Best Regards

Jacky 

Hi,Shankari

 1) SYSCLKOUT pin is DSP clock output. So is there any system clock pin to monitor ARM speed?

 2) For K2E , as I check the  datasheet. I believe the DEVSPEED address should been 0x02620C98, not 0x002620C90.

     If I have something wrong ,please correct me.

Thanks a lot

Best Regards

Jacky

Jacky,

Yes, you are right...

I had that question (2) posted to my other experts before even I post my answers here.

Let us hear from Mukul for both the questions.

Thanks and Regards

Shankari

Jacky,

Updated the gel file and attached here.

3683.K2H_register_dump_Updated.gel

Regards

Shankari

I believe for K2E there may not be any distinction on silicon revs , as we only have SR1.0. so 0x98 is the right base address. 

For K2H , you will need to confirm what silicon revision you have. Hopefully it is SR2.x. It is possible that you have an older EVM.

From SYSCLKOUT , it appears that that the devices are running at expected speed?

When you are running the benchmark, it is an ARM side benchmark or DSP side benchmark?

Let me know if i am missing something

Hello Jacky, 

I looked through the design specs for K2E and found that that the DEVSPEED field for K2E is on 0x02620C90.  The datasheet is incorrect in showing that as 0x02620C98.

From the register dump snapshot that was shared in  Jacky's post, the 0x02620C90 shows 0x00040004, so it is confirmed that you have a 1.25GHz part on K2E. This is also being validated by the SYSCLKOUT.

So hopefully that addresses/closes the issue of whether you have the right speed grade parts etc. 

There are no plans to update the datasheet at this point. 

If you are still running into performance differences on K2H and K2E , i recommend opening another thread to discuss more specifics of that.

Hope this helps.

Regards

Mukul 

Jacky,

Additionally, For K2E (DSP Speed --> 1.25 GHz), the same (SYSCLKOUT and DEVSPEED ) registers matches with the part number K2E IC : 66AK2E05XABD25 ( Part number you shared in your posts-above in reply to the PO units.) as well as with the device nomenclature.

 For K2H,  (ARM Speed -- > 1.4 GHz / DSP Speed --> 1.2 GHz ), the same matches with the device masking (1.2GHZ/1.4GHZ )

Regards

Shankari

Hi, Shankari

Even the DEVSPEED field for K2E is on 0x02620C90 and show  0x00040004, but form the datasheet I only find 1350MHz frequency , can not find 1250MHz.

Is it other error on K2E datasheet?

Best Regards

Jacky

Hi, Mukul

Mukul Bhatnagar said:

When you are running the benchmark, it is an ARM side benchmark or DSP side benchmark?

I confused about "benchmark".

How do I confirm that it is ARM or DSP side benchmark?

Best Regards

Jacky

Jacky

Yes there is likely the other error in the datasheet. 

The speed grades offered are the ones listed in section 3.4.2 of the datasheet -part number legend. 1.25GHz ARM/DSP or 1.4 GHz ARM/DSP 

The fields are 

By benchmark i meant whatever test you are running. In one of your older post you had mentioned the following 

 My team use same program to test the performance on both K2H and K2E IC. They find K2E is slower about 1.5s than K2H on every 28s.I want to confirm why K2E is slower?

Where is your team running this code, on ARM or DSP?

Hi Mukul

My team code run both ARM and DSP.

Let me tell you about  our design, we use only one ARM to run Linux system with our application.

One DSP control hardware and  communicate with ARM.

Jacky zhao said:

I check  SYSCLKOUT  output

            I could see K2H is running on 1.2GHz, K2E is running 1.25GHz

But I capture the waveform on only run Linux on ARM, I does not run DSP core.

Best Regards

Jacky

Jacky,

My suggestion is to close this post and open up a new post as the new one gets more attention than appending questions on the same post.

And also, this post has got appropriate answers for all your three questions of your first post.

Regards

Shankari

Hi,  Shankari and Mukul

Thanks for your strong support.

I will close this post.

Best Regards

Jacky