This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS570LC4357 vs TMS570LS3137

Other Parts Discussed in Thread: TMS570LS3137, TMS570LC4357, HALCOGEN

Good morning

We are currently using the device TMS570LS3137 and we have experienced with its strength and weakness points.

We saw the newest TMS570LC4357 and we are very interested to.

We would like to know, in order of priority:

1) Is TMS570LC4357 pin to pin compatible with TMS570LS3137?

2) With TMS570LC4357 is it possible to transimt Ethernet packets from EMIF SDRAM to EMAC at 100Mbit/s without to necessarily copy data to internal RAM before tx (as for TMS570LS3137)?

3) With TMS570LC4357 Are the LDM and STM Load and Store Multiple asm instructions working in EMIF SDRAM (bug inTMS570LS3137)?

4) When TMS570LC4357 will be available for sell, because it seems still under development, it's right?

Thank you in advance 

Paolo

  • Hi Paolo,

      We are glad that you are interested in our high performance LC4357 device.

      1. Yes, both device are pin compatible. As a matter of fact. the LC4357 and LS3137 shares the same HDK board. Of course, the LC4357 has more capabilities that some of the NC terminals in the LS3137 ZWT package will be populated with new functions. 

      2. I don't think this will change much between the two devices. The SDRAM is an external memory. If the EMAC buffer pointer points to the SDRAM then the EMAC DMA will have take some time to read them from the EMIF interface. 

      3. Yes, the STM bug in LS3137 does not apply to LC4357. Also FYI, the bug in LS3137 is fixed in the latest LS3137 silicon revision.

      4. LC4357 is available now and actually for a while. Please order your samples. 

  • Hi Charles

    Thank you for the answer.

    Another question: 

    We already have a PCB developed and working with our LS3137. Then,

    1) Is it possible to replace the LS3137 with LC4357, by leaving the newest pins as "Not Connected" (in order to not change the PCB master) ?

    2) Which are the consequences of this operation and the troubles to take into consideration?

    Thanks again

    Paolo

  • Paolo,
    Yes, you can replace LS3137 with LC4357 on your board. Bear in mind that they are pin compatible but not binary code compatible.
  • Thank you Charles

    This seems a good start point.

    1) Is it possible (i will check) to use the same oscillator 16MHz that i already have?

    2) Which Halcolgen version supports the newest LC4357?

    Paolo

  • HI Paolo,

    1. If the 16MHz OSC works on your 3137 device then it will work fine with LC4357.
    2. 04.05.02 is the latest HalCoGen version.
  • Hi, Charles.
    We ordered and received the samples part number TMP5704357BZWTQQ1.
    This is only a product/part number in PREVIEW status.
    We would like to know when the ACTIVE status will be available for production and the final part number.
    Best regards, Gianluca & Paolo.
  • Hi Gianluca,
    Just talked to the project manager who handles this device and he said the plan is September this year.
  • Hello Charles,

    On our previous products we used the TMS570LS3137. Now, on the new version we would like to increase the performances of our products. As the TMS570LC4357 is pin to pin compatible it can be a good candidate.

    According to the datasheet the ARM processor performances are equivalent 1.66DMIPS/MHz but the CortexR5F can be clocked up to 300MHz instead of 180MHz on the R4F. Is this due to a new technology used on the 4357 or is the technology the same on both components?

    I have some doubts on the performances that can be reached with the 4357 in comparison to the 3137. On one hand the ARM R5F clock can be up to 300MHz whereas the ARM R4F of the 3137 can be up to 180MHz. On the other hand the system clock frequency (HCLK) on the 4357 is limited at 150MHz whereas on the 3137 it can reach 180 MHz. Our products use some external SDRAM memory through the EMIF bus. It seems that on the 3137 the EMIF clock was limited to 50MHz whereas on the 4357 it can reach 100MHz, am I right?

    Do you have any results about the performances (number of DMIPS) of the TMS570LC4357ZWT in the following configuration: ?
    - .code from the internal Flash,
    - heap and stack on internal SRAM
    - data on the external SDRAM
    - GCLK=300MHz; HCLK=150MHz and VCLK3=75MHz

    Thank you in adavance,
    Regards,

    Mario.
  • Hi Mario,

      The performance improvement on LC4357 comes from the architecture, not technology. Both LC4357 and LS3137 are built off of the same 65nm technology. However, the LC4357 is a cache based architecture which allows it to run at much higher CPU frequency than LS3137. The LS3137 uses tightly couple memory architecture which limits its operating frequency to 180MHz. Since the LC4357 CPU frequency runs at 300MHz the rest of the system can not keep up with it at the same frequency. This is the reason the system clock HCLK is spec'ed at 150MHz. As long as the CPU has cache hit on the instructions and data, it is running at maximum performance with zero wait state.

      Yes, the EMIF on LC4357 is spec'ed to 100MHz for SDRAM interface.

      

  • HI Mario,
    If your question answered is answered can you please click the 'Verify Answer' button to close the thread? Thanks.