目标是用FPGA来回传送好大小的数据块。
我注意到的第一件事是外部存储器接口组件。他们看起来很理想,直到我注意到他们的公共汽车钟最大是33兆赫。
因为在FPGA方面有很大的灵活性,所以我决定我可以“滚我自己”的界面。这将避免拖垮整个总线时钟,并可能使时钟与其他模块混淆。我认为16位FIFOIN和FIFOUT与DMA和一点胶水逻辑将是解决方案。不幸的是,我没有意识到获得一个工作的先进先出的范围。
然后,我认为8位FIFOIN和8位控制寄存器与DMA可能是足够的。DMA效率低,但很可能仍然能完成任务。在这种情况下,我不知道什么时候用新的字节更新控制寄存器,所以我不能得到一个可用的写信号到FPGA。
以上来自于百度翻译
以下为原文
The goal is to transfer good size blocks of data back and forth with an FPGA.
The first thing I noticed was the external memory interface components. They seemed ideal until I noticed their bus clock maximum is 33Mhz.
Since there is great flexibility on the FPGA side, I decided I could "roll my own" interface. This would avoid dragging down the whole bus clock and possibly messing with the clocks to my other modules. I thought 16 bit FIFOin and FIFOout with DMA and a bit of glue logic would be the solution. Unfortunately, I didn't realized the scope of getting a working FIFOout.
Then I thought 8 bit FIFOin and 8 bit control register out with DMA might be adequate. The DMA would be less efficient, but would probably still get the job done. In this scenario, I don't know when the control register is updated with a new byte, so I can't get a usable write signal to the FPGA.
目标是用FPGA来回传送好大小的数据块。
我注意到的第一件事是外部存储器接口组件。他们看起来很理想,直到我注意到他们的公共汽车钟最大是33兆赫。
因为在FPGA方面有很大的灵活性,所以我决定我可以“滚我自己”的界面。这将避免拖垮整个总线时钟,并可能使时钟与其他模块混淆。我认为16位FIFOIN和FIFOUT与DMA和一点胶水逻辑将是解决方案。不幸的是,我没有意识到获得一个工作的先进先出的范围。
然后,我认为8位FIFOIN和8位控制寄存器与DMA可能是足够的。DMA效率低,但很可能仍然能完成任务。在这种情况下,我不知道什么时候用新的字节更新控制寄存器,所以我不能得到一个可用的写信号到FPGA。
以上来自于百度翻译
以下为原文
The goal is to transfer good size blocks of data back and forth with an FPGA.
The first thing I noticed was the external memory interface components. They seemed ideal until I noticed their bus clock maximum is 33Mhz.
Since there is great flexibility on the FPGA side, I decided I could "roll my own" interface. This would avoid dragging down the whole bus clock and possibly messing with the clocks to my other modules. I thought 16 bit FIFOin and FIFOout with DMA and a bit of glue logic would be the solution. Unfortunately, I didn't realized the scope of getting a working FIFOout.
Then I thought 8 bit FIFOin and 8 bit control register out with DMA might be adequate. The DMA would be less efficient, but would probably still get the job done. In this scenario, I don't know when the control register is updated with a new byte, so I can't get a usable write signal to the FPGA.
举报