Agilex 5 Premium Dev Kit Ethernet Performance
Hello! We built the golden sample image following the HPS GSRD User Guide with additional packages to profile/evaluate the board and experience performance problems when sending data over ethernet. The test setup is a host connected to the dev kit and sending data to test the throughput. First, we used iperf3 with zero copy flag, which caps at about 940 Mbit/s with almost no variation. Without zero copy, iperf3 caps at about ~880 Mbit/s with some variation down to 629 Mbit/s, see attachment 1.png. With our custom application that also does some additional work, we’d expect about 430-440 Mbit/s, but cap at about 300 Mbit/s, with lots of time spent in kernel again, see attachment 2.png. From the first investigation, we suspect the driver can’t keep up with the generated data and can’t send it fast enough to the host. We are wondering whether we can adjust something in the kernel (driver) or in the image so that we can improve the throughput with heavy workloads. Kind regards!
Support Request to Debug Signal Tap Issue with SoC Based Design(GHRD) on Agilex-7 FPGA
Hi Altera Support Team, Greetings from Logic Fruit Technologies. This is Bhanu Pratap and I am working on the project in which we are using Agilex-7 SOC FPGA. Goal: Utilize Agiliex-HPS's capability on a Linux OS to control dynamic features of IP Cores at runtime. Problem Statement: We are facing an issue detecting SignalTap with SoC-based designs for runtime debugging while it's enabled and added from the Quartus project to generate the programming file. Brief Description: We are using Quartus 2025.1.1 and Altera's Agilex-7 Devkit(DKSIAAGI027FA: Agilex™ 7 FPGA I-Series Transceiver-SoC Development Kit -> 4x F-Tile, Rev C). We downloaded the base GHRD design from Altera's GitHub webpage for the corresponding Quartus version. We are following each step to test GHRD and build generation for the Linux OS. We are using HPS First booting mode, we generated the build without modification and tested it. It's working fine according to the GSRD user guide. We have added a custom AXI4 Slave based CSR module to the HSP Lightweight bus to test basic register read/write using Platform Designer. We regenerated the build and re-tested but we are not getting the expected results so we added SignalTap for debugging. Build is generated using the STP file, programmed, and the OS is booting but it's not detecting SignalTap. This looks more like a process gap. We seek your support to fix this issue. Note: If we use an FPGA design without HPS, we are able to add and detect SignalTap. Seeking for urgent support. Thanks Bhanu PratapTimings eMMC
Hi, latest datasheet of SoC « 813918/2026.01.05 » lists in paragraph « HPS SD/eMMC Timing Characteristics » the constraints of eMMC, applicable to memory component but does not define any timing data in input/output of SoC. There is no Tco min/max of CMD/DATA at SoC output, as well as Tsetup/hold of CMD/DATA at SoC input. Can you provide Tco and Tsetup/hold for eMMC usage (Legacy, HS_SDR, HS_DDR, HS200, HS400) ? thanks
S10 hps fpga2sdram bridge low speed
Hello, i have some problems I have a project with stratix 10 with hps I need to use ddr4 with hps, so I enabled 3 fpga2sdram bridges with 128 bit width Via u-boot smc configured and enabled them, but measured speed is not enough. When I use onle one bridge my speed is approximately 32 gbit/s (bridge and master frequency 350 MHz) But when I use all 3 bridges it becomes 20 gbit/s I use avmm bridge to connect to axi fpga2sdram bridge with 128 bits width, max pending writes 16, max burst size 128, use only write to bridge ECC in emif (hmc) is disabled I tried to use QoS for bridges, set them to bypass P.S. If i use the same board with firmware without hps, i get 130 gbit/s (with disabled hps) Quartus Pro 21.4 Any help would be useful!
Agilex 5 HPS EMAC2 SWR bit in DMA_Mode Register Not Clearing
Hello, Following section 5.1.7 of the Agilex 5 Hard Processor System Technical Reference Manual, and based on what I found in U-Boot and Linux, I am writing an Ethernet driver that uses EMAC2 with an RGMII interface. I am currently stuck at the software reset using the SWR bit of the DMA_Mode register, as it remains set to 1. Our boot flow on the AXE5-Eagle board is ATF -> U-Boot -> our OS. We use the U-Boot from https://github.com/ArrowElectronics/u-boot-socfpga/tree/socfpga_v2025.07 with some adaptations to load our OS instead of Linux. When I read the DMA_Mode register before performing any EMAC setup in our OS, I already see the value set to 1. After bringing the PHY out of reset and setting the SWR bit again, it does not clear. I also tried putting the EMAC into reset using the per0modrst register, setting the System Manager tsn2 register to 1 again, and then bringing the EMAC out of reset. This did not help, and after resetting the EMAC I can no longer read PHY registers over MDIO. When I load Linux instead of our OS, I can see that the reset using the SWR bit works correctly, but I can’t find what is missing in my implementation. Any help would be greatly appreciated. Thanks, Sarah For additional context, I read the following values before putting the EMAC2 in reset: PHY ID1: 0x283 Main PLLGLOB: 35000103 Per PLLGLOB: 35000103 DMA_Mode: 00000001 SYSMGR_TSN2: 1 EMAC2_DWCXG_CORE_MAC_TX_CONF: 0 CLKMGR_PERPLL_EN: fffffff CLKMGR_PERPLL_BYPASS: 0 CLKMGR_PERPLL_EMACCTL: 0 CLKMGR_CTLGRP_EMACACTR: 10001 CLKMGR_CTLGRP_EMACPTPCTR: 0 CLKMGR_CTLGRP_EXTCNTRST: 0Agilex 5 HPS EMAC2 SWR bit in DMA_Mode Register Not Clearing
Hello, Following section 5.1.7 of the Agilex 5 Hard Processor System Technical Reference Manual, and based on what I found in U-Boot and Linux, I am writing an Ethernet driver that uses EMAC2 with an RGMII interface. I am currently stuck at the software reset using the SWR bit of the DMA_Mode register, as it remains set to 1. Our boot flow on the AXE5-Eagle board is ATF -> U-Boot -> our OS. We use the U-Boot from https://github.com/ArrowElectronics/u-boot-socfpga/tree/socfpga_v2025.07 with some adaptations to load our OS instead of Linux. When I read the DMA_Mode register before performing any EMAC setup in our OS, I already see the value set to 1. After bringing the PHY out of reset and setting the SWR bit again, it does not clear. I also tried putting the EMAC into reset using the per0modrst register, setting the System Manager tsn2 register to 1 again, and then bringing the EMAC out of reset. This did not help, and after doing this I can no longer read PHY registers over MDIO. When I load Linux instead of our OS, I can see that the reset using the SWR bit works correctly, but I can’t find what is missing in my implementation. Any help would be greatly appreciated. Thanks, Sarah For additional context, I read the following values before putting the EMAC2 in reset: PHY ID1: 0x283 Main PLLGLOB: 35000103 Per PLLGLOB: 35000103 DMA_Mode: 00000001 SYSMGR_TSN2: 1 EMAC2_DWCXG_CORE_MAC_TX_CONF: 0 CLKMGR_PERPLL_EN: fffffff CLKMGR_PERPLL_BYPASS: 0 CLKMGR_PERPLL_EMACCTL: 0 CLKMGR_CTLGRP_EMACACTR: 10001 CLKMGR_CTLGRP_EMACPTPCTR: 0 CLKMGR_CTLGRP_EXTCNTRST: 0
Why does the system report an error when generating rbf from sof files and fsbl files?
Error message: Error: Internal Error: Sub-system: BITASM, File: /quartus/pgm/bitasm/bitasm_common_code.cpp, Line: 518 HPS data start address(-1950584) is not 16 aligned Device and tool information： The device used is Stratix 10 1SX110HN2F43I2VG, without using the Stratix 10 SoC Development Kit； Quartus Prime Pro25.1.1 U-boot source code：u-boot-socfpga-socfpga_v2025.04 ATF source code：arm-trusted-firmware-socfpga_v2.13.0 Operation steps： Simplified the Platform Designer section of the Stratix 10 GHRD project； 【Device and Pin Options】->【Configuration】Set the HPS/FPGA configuration order to be HPS First； The Quartus full compilation generates the sof file in the "output_files" directory； Compile the ATF source code, and obtain the bl31.bin file in the path of ./build/stratix10/release； Copy the bl31.bin file to the root directory of u-boot, compile the u-boot source code, and obtain the u-boot-spl file in the ./spl/ directory； Convert u-boot-spl to u-boot-spl.hex and copy it to the output_files directory； Open the Programming File Generator tool and configure the Output Files： Configure Input Files, add sof and HEX files： 9. Configuration Device： 10. Generate error：
Agilex 5 EMAC GMII loopthrough: signals are not toggling in Fabric
I am trying to route GMII signals through the Agilex5 HPS to the Fabric. In the .dts I am using, I have the following settings for the two EMACs: &gmac0 { status = "okay"; mac-mode = "gmii"; phy-mode = "gmii"; // must be added, and if no PHY, then add fixed link //phy-handle = <&emac0_phy0>; max-frame-size = <9000>; fixed-link { speed = <1000>; full-duplex; }; mdio0 { #address-cells = <1>; #size-cells = <0>; compatible = "snps,dwmac-mdio"; emac0_phy0: ethernet-phy@0 { reg = <0>; }; }; }; &gmac1 { status = "okay"; mac-mode = "gmii"; phy-mode = "gmii"; //phy-handle = <&emac1_phy0>; max-frame-size = <9000>; fixed-link { speed = <1000>; full-duplex; }; mdio0 { #address-cells = <1>; #size-cells = <0>; compatible = "snps,dwmac-mdio"; emac1_phy0: ethernet-phy@0 { reg = <0>; }; }; }; In the RTL I am connecting the two EMACs as followed: logic user_clk_pll_125, user_clk_pll_25, user_clk_pll_2_5; // 1G / 100M / 10M (* noprune *) logic [7:0] tx_dummy_counter; (* noprune *) logic emac0_mac_tx_clk_o_wire, emac0_mac_tx_clk_i_wire, emac0_mac_rx_clk_wire, emac0_mac_rst_tx_n_wire, emac0_mac_rst_rx_n_wire; (* noprune *) logic emac0_mac_txen_wire, emac0_mac_txer_wire, emac0_mac_rxdv_wire, emac0_mac_rxer_wire, emac0_mac_col_wire, emac0_mac_crs_wire; (* noprune *) logic [7:0] emac0_mac_rxd_wire; (* noprune *) logic [2:0] emac0_mac_speed_wire; (* noprune *) logic [7:0] emac0_mac_txd_o_wire; (* noprune *) logic [7:0] rx_dummy_counter; (* noprune *) logic emac1_mac_tx_clk_o_wire, emac1_mac_tx_clk_i_wire, emac1_mac_rx_clk_wire, emac1_mac_rst_tx_n_wire, emac1_mac_rst_rx_n_wire; (* noprune *) logic emac1_mac_txen_wire, emac1_mac_txer_wire, emac1_mac_rxdv_wire, emac1_mac_rxer_wire, emac1_mac_col_wire, emac1_mac_crs_wire; (* noprune *) logic [7:0] emac1_mac_rxd_wire; (* noprune *) logic [2:0] emac1_mac_speed_wire; (* noprune *) logic [7:0] emac1_mac_txd_o_wire; assign emac0_mac_rx_clk_wire = emac1_mac_tx_clk_o_wire; // 1G assign emac1_mac_rxdv_wire = emac0_mac_txen_wire; assign emac1_mac_rxer_wire = emac0_mac_txer_wire; assign emac1_mac_rxd_wire = emac0_mac_txd_o_wire; assign emac1_mac_col_wire = 1'b0; assign emac1_mac_crs_wire = 1'b0; assign emac0_mac_rx_clk_wire = emac0_mac_tx_clk_o_wire; // 1G assign emac0_mac_rxdv_wire = emac1_mac_txen_wire; assign emac0_mac_rxer_wire = emac1_mac_txer_wire; assign emac0_mac_rxd_wire = emac1_mac_txd_o_wire; assign emac0_mac_col_wire = 1'b0; assign emac0_mac_crs_wire = 1'b0; The GMII signals are exported from the Agilex HPS as followed: When the system boots, the following can be seen is dmesg: [ 1.443647] socfpga-dwmac 10810000.ethernet: Adding to iommu group 0 [ 1.450679] socfpga-dwmac 10810000.ethernet: IRQ eth_wake_irq not found [ 1.457291] socfpga-dwmac 10810000.ethernet: IRQ eth_lpi not found [ 1.463542] socfpga-dwmac 10810000.ethernet: RX VLAN HW Stripping [ 1.469741] socfpga-dwmac 10810000.ethernet: SMTG Hub Cross Timestamp supported [ 1.477398] socfpga-dwmac 10810000.ethernet: User ID: 0x76, Synopsys ID: 0x31 [ 1.484534] socfpga-dwmac 10810000.ethernet: XGMAC2 [ 1.489489] socfpga-dwmac 10810000.ethernet: DMA HW capability register supported [ 1.496943] socfpga-dwmac 10810000.ethernet: RX Checksum Offload Engine supported [ 1.504396] socfpga-dwmac 10810000.ethernet: COE Type 1 [ 1.509603] socfpga-dwmac 10810000.ethernet: TX Checksum insertion supported [ 1.516623] socfpga-dwmac 10810000.ethernet: TSO supported [ 1.522089] socfpga-dwmac 10810000.ethernet: Enable RX Mitigation via HW Watchdog Timer [ 1.530076] socfpga-dwmac 10810000.ethernet: device MAC address 42:ca:f5:1e:55:80 [ 1.537533] socfpga-dwmac 10810000.ethernet: Enabled L3L4 Flow TC (entries=16) [ 1.544737] socfpga-dwmac 10810000.ethernet: Enabled RFS Flow TC (entries=10) [ 1.551847] socfpga-dwmac 10810000.ethernet: TSO feature enabled [ 1.557831] socfpga-dwmac 10810000.ethernet: SPH feature enabled [ 1.563815] socfpga-dwmac 10810000.ethernet: TX COE limited to 2 tx queues [ 1.570665] socfpga-dwmac 10810000.ethernet: Using 40/40 bits DMA host/device width [ 1.581335] socfpga-dwmac 10820000.ethernet: Adding to iommu group 1 [ 1.588338] socfpga-dwmac 10820000.ethernet: IRQ eth_wake_irq not found [ 1.594945] socfpga-dwmac 10820000.ethernet: IRQ eth_lpi not found [ 1.601179] socfpga-dwmac 10820000.ethernet: RX VLAN HW Stripping [ 1.607380] socfpga-dwmac 10820000.ethernet: SMTG Hub Cross Timestamp supported [ 1.614905] socfpga-dwmac 10820000.ethernet: User ID: 0x76, Synopsys ID: 0x31 [ 1.622027] socfpga-dwmac 10820000.ethernet: XGMAC2 [ 1.626982] socfpga-dwmac 10820000.ethernet: DMA HW capability register supported [ 1.634436] socfpga-dwmac 10820000.ethernet: RX Checksum Offload Engine supported [ 1.641890] socfpga-dwmac 10820000.ethernet: COE Type 1 [ 1.647097] socfpga-dwmac 10820000.ethernet: TX Checksum insertion supported [ 1.654117] socfpga-dwmac 10820000.ethernet: TSO supported [ 1.659583] socfpga-dwmac 10820000.ethernet: Enable RX Mitigation via HW Watchdog Timer [ 1.667568] socfpga-dwmac 10820000.ethernet: device MAC address 3e:47:0a:4f:7b:96 [ 1.675024] socfpga-dwmac 10820000.ethernet: Enabled L3L4 Flow TC (entries=16) [ 1.682221] socfpga-dwmac 10820000.ethernet: Enabled RFS Flow TC (entries=10) [ 1.689330] socfpga-dwmac 10820000.ethernet: TSO feature enabled [ 1.695314] socfpga-dwmac 10820000.ethernet: SPH feature enabled [ 1.701298] socfpga-dwmac 10820000.ethernet: TX COE limited to 2 tx queues [ 1.708147] socfpga-dwmac 10820000.ethernet: Using 40/40 bits DMA host/device width [ 1.718293] socfpga-dwmac 10830000.ethernet: Adding to iommu group 2 [ 1.725245] socfpga-dwmac 10830000.ethernet: IRQ eth_wake_irq not found [ 1.731850] socfpga-dwmac 10830000.ethernet: IRQ eth_lpi not found [ 1.738075] socfpga-dwmac 10830000.ethernet: RX VLAN HW Stripping [ 1.744242] socfpga-dwmac 10830000.ethernet: SMTG Hub Cross Timestamp supported [ 1.751730] socfpga-dwmac 10830000.ethernet: User ID: 0x76, Synopsys ID: 0x31 [ 1.758854] socfpga-dwmac 10830000.ethernet: XGMAC2 [ 1.763807] socfpga-dwmac 10830000.ethernet: DMA HW capability register supported [ 1.771261] socfpga-dwmac 10830000.ethernet: RX Checksum Offload Engine supported [ 1.778712] socfpga-dwmac 10830000.ethernet: COE Type 1 [ 1.783917] socfpga-dwmac 10830000.ethernet: TX Checksum insertion supported [ 1.790936] socfpga-dwmac 10830000.ethernet: TSO supported [ 1.796400] socfpga-dwmac 10830000.ethernet: Enable RX Mitigation via HW Watchdog Timer [ 1.804372] socfpga-dwmac 10830000.ethernet: Enabled L3L4 Flow TC (entries=16) [ 1.811565] socfpga-dwmac 10830000.ethernet: Enabled RFS Flow TC (entries=10) [ 1.818673] socfpga-dwmac 10830000.ethernet: TSO feature enabled [ 1.824656] socfpga-dwmac 10830000.ethernet: SPH feature enabled [ 1.830639] socfpga-dwmac 10830000.ethernet: TX COE limited to 2 tx queues [ 1.837487] socfpga-dwmac 10830000.ethernet: Using 40/40 bits DMA host/device width ... [ 11.730428] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-0 [ 11.747001] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-1 [ 11.770275] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-2 [ 11.778338] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-3 [ 11.782322] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-4 [ 11.787412] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-5 [ 11.790998] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-6 [ 11.796018] socfpga-dwmac 10830000.ethernet eth2: Register MEM_TYPE_PAGE_POOL RxQ-7 [ 11.813781] fpga_manager fpga0: Stratix10 SOC FPGA Manager registered [ 11.913063] socfpga-dwmac 10830000.ethernet eth2: PHY [stmmac-2:01] driver [Micrel KSZ9031 Gigabit PHY] (irq=POLL) [ 11.913393] socfpga-dwmac 10830000.ethernet eth2: No Safety Features support found [ 11.913478] socfpga-dwmac 10830000.ethernet eth2: IEEE 1588-2008 Advanced Timestamp supported [ 12.001908] socfpga-dwmac 10830000.ethernet eth2: registered PTP clock [ 12.007839] socfpga-dwmac 10830000.ethernet eth2: FPE workqueue start [ 12.007940] socfpga-dwmac 10830000.ethernet eth2: configuring for phy/rgmii-id link mode [ 12.155854] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-0 [ 12.159161] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-1 [ 12.180998] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-2 [ 12.191086] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-3 [ 12.199787] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-4 [ 12.208205] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-5 [ 12.218464] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-6 [ 12.229854] socfpga-dwmac 10820000.ethernet eth1: Register MEM_TYPE_PAGE_POOL RxQ-7 [ 12.247722] socfpga-dwmac 10820000.ethernet eth1: No Safety Features support found [ 12.247967] socfpga-dwmac 10820000.ethernet eth1: IEEE 1588-2008 Advanced Timestamp supported [ 12.331440] socfpga-dwmac 10820000.ethernet eth1: registered PTP clock [ 12.332900] socfpga-dwmac 10820000.ethernet eth1: FPE workqueue start [ 12.332987] socfpga-dwmac 10820000.ethernet eth1: configuring for fixed/gmii link mode [ 12.343803] socfpga-dwmac 10820000.ethernet eth1: Link is Up - 100Mbps/Full - flow control off [ 12.344209] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready [ 12.449807] of-fpga-region soc:base_fpga_region: FPGA Region probed [ 12.469882] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-0 [ 12.479649] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-1 [ 12.482568] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-2 [ 12.490255] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-3 [ 12.493314] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-4 [ 12.501735] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-5 [ 12.514200] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-6 [ 12.536765] socfpga-dwmac 10810000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-7 [ 12.582718] socfpga-dwmac 10810000.ethernet eth0: No Safety Features support found [ 12.583005] socfpga-dwmac 10810000.ethernet eth0: IEEE 1588-2008 Advanced Timestamp supported [ 12.591157] socfpga-dwmac 10810000.ethernet eth0: registered PTP clock [ 12.601012] socfpga-dwmac 10810000.ethernet eth0: FPE workqueue start [ 12.601201] socfpga-dwmac 10810000.ethernet eth0: configuring for fixed/gmii link mode [ 12.611650] socfpga-dwmac 10810000.ethernet eth0: Link is Up - 100Mbps/Full - flow control off [ 12.615095] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready ... (this log was from testing 100M, but the same occurs for 1G) In signal tap, no wires are toggled (even though it seems the output clocks are fine) Please help!Cyclone V HPS FPGA2SDRAM Clock Queries
Dear Intel and all, Having some very puzzling behavior on HPS SDRAM and FPGA fabric bridge. If a fast clock i.e. 148MHz is running on 128bit AXI3 aka f2h_sdram0 mostly read action. And another AXI3 is using the remain 128bit bus with 144MHz aka f2h_sdram1 mostly on write. As such the system will stuck on distro aka Linux. With all these background could engineer or internal stuffs help. What is the restriction or constraints to use these bus under safe and stable speed? Forgot to provide stable situation: If the write dominated bus is reduced to 100MHz then the system is stable and no stall is found. So this makes a very strong feeling that the write cache is having issue? maybe CMA insufficient? Brian
