UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Mentor xilinxacct
Mentor
252 Views
Registered: ‎10-23-2018

how to avoid HLS ARRAY_PARTITION stall potential?

Jump to solution

If you have an array and you want to process the first half and last half in parallel, you can use the pragma HLS ARRAY_PARTITION variable=buf block factor=2

But, this by nature, give the potential for a stall at the buf[1] position

other factors up to the ‘complete’ introduce more potential (e.g factor 4 would have potential at s[2] & s[3])… complete avoids this by implementing FIFO, but this is a large expansion in utilization, and overkill for the need.

Is there a different pragma to use to avoid the stall potential OR parameter on this pragma to do likewise OR some method to remove the potential?

Note: cyclic is not an option for this particular case, as the order of the items in the buffer are significant.

Thanks

Tags (2)
0 Kudos
1 Solution

Accepted Solutions
Scholar u4223374
Scholar
212 Views
Registered: ‎04-26-2015

Re: how to avoid HLS ARRAY_PARTITION stall potential?

Jump to solution

I'm not sure how this can cause a stall. I've regularly used arrays that are effectively block partitioned and have never had HLS actually stall while accessing them. You can obviously hit dependency constraints, but that's a hardware limitation rather than a software one.

2 Replies
Scholar u4223374
Scholar
213 Views
Registered: ‎04-26-2015

Re: how to avoid HLS ARRAY_PARTITION stall potential?

Jump to solution

I'm not sure how this can cause a stall. I've regularly used arrays that are effectively block partitioned and have never had HLS actually stall while accessing them. You can obviously hit dependency constraints, but that's a hardware limitation rather than a software one.

Mentor xilinxacct
Mentor
202 Views
Registered: ‎10-23-2018

Re: how to avoid HLS ARRAY_PARTITION stall potential?

Jump to solution

@u4223374

That's what I thought too... Glad you confirmed it is the best it can be, and I can ignore the warning and know there is no room for a more optimal pattern. 

Thanks again

0 Kudos