完善资料让更多小伙伴认识你,还能领取20积分哦, 立即完善>
我有一个SX475T,我已经对我设计的很大一部分进行了布局规划,但似乎我无法通过芯片的右下角以及右上角。
我有一个逻辑实例化了3次。 我在右上方放了一个pblock,垂直覆盖2个时钟区域。 下面是接下来的2个时钟区域中的第2个pblock(与第一个完全相同的逻辑)。 这正确地路由和解析。 然后我添加第三个副本并将其放在下面,并且任何一个地图都会失败,说没有空格或者标准杆具有数百万的计时分数。 我知道有足够的空间,因为其他2个pblock具有与源相同的NGC文件并且适合相同大小的块,。 我已经做了第三个pblock来覆盖所有剩余的时钟区域(大小加倍),它仍然说它不适合。 驱动和馈电的逻辑是在芯片的右侧中心,因此它应该到达底部以及顶部(相同的路径长度)。 物理资源估计表明,对于pblock,thsi pblock使用67%lut,52%ff(当我将pblock大小加倍时为1/2) 有没有一个隐藏的硅原因,右下角将更难以路由芯片的其余部分? 我认为xilinx曾经说过路由和右路是首选,但我不知道为什么,或者如果这仍然是真的。 以上来自于谷歌翻译 以下为原文 I have a SX475T for which I have floorplanned a large part of my design but it appears I cant route to the bottom right corner of the chip as well as the top right. I have one piece of logic that is instantiated 3 times. I put a pblock in the upper right that covers 2 clock regions vertically. Below it is the 2nd pblock (exactly same logic as first) in the next 2 clockregions. This routes and pars correctly. Then I add the 3rd copy and place it below that and either map fails saying there isn't space or par has a timing score in the millions. I know there is enough space because the other 2 pblocks have the same NGC file as source and fit in the same size block, . I have go as far as making the 3rd pblock to cover all the remaining clock regions (doubling the size) and it still says it wont fit. The logic that drives and feeds this is in the right left side of the chip centered so it should reach the bottom just as well as the top (same path length). The physical resource estimation says thsi pblock uses 67%lut, 52%ff for the pblocks (1/2 that when I doubled the pblock size) Is there a hidden silicon reason that the bottom right would be harder to route that the rest of the chip? I think xilinx used to say routing up and right was preferred but I don't know why, or if that's still true. |
|
相关推荐
3个回答
|
|
如果您遇到拟合失败,则生成的Place:543错误将列出无法适合的组件。
这将提供对故障根本原因的深入了解。 涉及哪些组件类型以及它们有什么共同之处? 以上来自于谷歌翻译 以下为原文 If you are getting a fitting failure, the resulting Place:543 error will list the components that failed to fit. This will provide insight into the root cause of the failure. What component types are involved and what do they have in common? |
|
|
|
我们一直在日志中上下,他们不会在信息方面提供太多帮助。
它们似乎在每次运行时显示随机信号,与一组一致的信号相对应。 它们似乎只是它与地方相关的最后一个。 它是寄存器,但正如我所说的那样,pblock中有足够的寄存器,因为它只是它自己的PBLOCK中同一个NGC文件的实例化。 在极端情况下,当我们使pblock尽可能大时,资源估算器显示23%的利用率。 是否有关于路由资源架构方式的文档? 我想知道你是否使用垂直路由资源它是在整个列上还是仅在时钟区域中消耗一个垂直路由路径? V6上的数据表对路由信息略有不足。 planahead中的芯片图显示了MAC核心中间的空白区域(我认为那就是生活在那里)。 这是真正芯片的一个很好的代表吗? 如果我有从左侧到右侧的信号,它们是否必须挤过这个黑色区域的上方和下方,如图中所示? 以上来自于谷歌翻译 以下为原文 We have been up and down the logs and they don't lend much in the way of info. They appear to show random signals each time it's run as apposed to a consistent set of signals. They appear to be just the last ones it tied to place. It's registers but as I said there are enough registers in the pblock since it's just an instantiation of the same NGC file in it's own PBLOCK. In the extremes when we make the pblock as big as possible the resource estimator shows 23% utilization. Is there any documentation on the way routing resource are architecture? I'm wondering if you use vertical routing resource does it consume one vertical routing path on the whole column or just in a clock region? The datasheet on the V6 is a little bare on routing information. The chip diagram in planahead shows a blank area in the middle where the MAC cores are (I think that's what lives there). Is that a good representation of the real chip? If I have signals that go from the left to the right side do they have to squeeze through above and below this black area as it would appear on the diagram? |
|
|
|
您的注册配件问题与一般路由资源无关。
这是与可用站点相关的打包/放置问题。 检查地图报告(.mrp)的区域组利用率部分,以检查有关该pblock的寄存器利用率的假设。 如果它很低那么拟合问题是由于除了整体利用之外的一些其他约束。 听起来你可能有控制装置问题干扰包装密度。 涉及什么样的时钟? 检查限制放置选项的时钟约束。 检查.par文件中的大型本地时钟。 最后有一张表汇总了所有时钟网络,包括本地时钟。 以上来自于谷歌翻译 以下为原文 Your register fitting issue is not related to general routing resources. It's a pack/place issue related to available sites. Check the area group utilization section of the map report (.mrp) to check your assumptions about register utilization for that pblock. If it is low then the fitting issue is due to some other constraint than overall utilization. It sounds like you might have control set issues interfering with packing density. What sort of clocks are involved? Check for clocking constraints that limit your placement options. Check for large local clocks in the .par file. There's a table at the end that summarizes all of the clock nets including local clocks. |
|
|
|
只有小组成员才能发言,加入小组>>
2429 浏览 7 评论
2831 浏览 4 评论
Spartan 3-AN时钟和VHDL让ISE合成时出现错误该怎么办?
2298 浏览 9 评论
3378 浏览 0 评论
如何在RTL或xilinx spartan fpga的约束文件中插入1.56ns延迟缓冲区?
2468 浏览 15 评论
有输入,但是LVDS_25的FPGA内部接收不到数据,为什么?
1332浏览 1评论
请问vc707的电源线是如何连接的,我这边可能出现了缺失元件的情况导致无法供电
595浏览 1评论
求一块XILINX开发板KC705,VC707,KC105和KCU1500
457浏览 1评论
2012浏览 0评论
737浏览 0评论
小黑屋| 手机版| Archiver| 电子发烧友 ( 湘ICP备2023018690号 )
GMT+8, 2024-12-28 08:49 , Processed in 1.309660 second(s), Total 80, Slave 64 queries .
Powered by 电子发烧友网
© 2015 bbs.elecfans.com
关注我们的微信
下载发烧友APP
电子发烧友观察
版权所有 © 湖南华秋数字科技有限公司
电子发烧友 (威廉希尔官方网站 图) 湘公网安备 43011202000918 号 电信与信息服务业务经营许可证:合字B2-20210191 工商网监 湘ICP备2023018690号