(I have one of these, it appears to be two chiplets, with half the cache disabled. Setting the BIOS to have a NUMA domain per LLC results in four numa domains. Also, 'lscpu' reports 4 L3 'instances') |
|||
(10 intermediate revisions by 3 users not shown) | |||
Line 1: | Line 1: | ||
− | {{amd title|EPYC | + | {{amd title|EPYC 7232P}} |
{{chip | {{chip | ||
− | |name=EPYC | + | |name=EPYC 7232P |
|no image=Yes | |no image=Yes | ||
|designer=AMD | |designer=AMD | ||
|manufacturer=TSMC | |manufacturer=TSMC | ||
|manufacturer 2=GlobalFoundries | |manufacturer 2=GlobalFoundries | ||
− | |model number= | + | |model number=7232P |
|part number=100-000000081 | |part number=100-000000081 | ||
+ | |part number 2=100-100000081WOF | ||
|market=Server | |market=Server | ||
|first announced=August 7, 2019 | |first announced=August 7, 2019 | ||
|first launched=August 7, 2019 | |first launched=August 7, 2019 | ||
+ | |release price=$450 | ||
|family=EPYC | |family=EPYC | ||
|series=7002 | |series=7002 | ||
|locked=Yes | |locked=Yes | ||
− | |frequency= | + | |frequency=3,100 MHz |
|turbo frequency=3,200 MHz | |turbo frequency=3,200 MHz | ||
− | |clock multiplier= | + | |clock multiplier=31 |
|isa=x86-64 | |isa=x86-64 | ||
|isa family=x86 | |isa family=x86 | ||
Line 22: | Line 24: | ||
|core name=Rome | |core name=Rome | ||
|core family=23 | |core family=23 | ||
+ | |core model=49 | ||
+ | |core stepping=B0 | ||
|process=7 nm | |process=7 nm | ||
|process 2=14 nm | |process 2=14 nm | ||
|technology=CMOS | |technology=CMOS | ||
|mcp=Yes | |mcp=Yes | ||
− | |die count= | + | |die count=2 |
|word size=64 bit | |word size=64 bit | ||
|core count=8 | |core count=8 | ||
Line 35: | Line 39: | ||
|package name 1=amd,socket_sp3 | |package name 1=amd,socket_sp3 | ||
}} | }} | ||
− | '''EPYC | + | '''EPYC 7232P''' is a {{arch|64}} [[octa-core]] [[x86]] server microprocessor designed and introduced by [[AMD]] in mid-[[2019]]. This [[multi-chip package|multi-chip processor]], which is based on the {{amd|Zen 2|Zen 2 microarchitecture|l=arch}}, incorporates logic fabricated [[TSMC]] [[7 nm process]] and I/O fabricated on [[GlobalFoundries]] [[14 nm process]]. The 7232P has a TDP of 120 W with a base frequency of 3.1 GHz and a {{amd|precision boost|boost}} frequency of up to 3.2 GHz. This processor supports single-socket configurations only and up to 4 TiB of memory per socket. |
+ | |||
+ | This processor has 8 CPU cores and 32 MiB L3 cache. A single Core Complex Die (plus I/O die) can provide these resources, however there seems to be no evidence that AMD produces any Zen 2-based EPYC or Threadripper processors with an odd number of CCDs, and the advertised maximum memory bandwidth appears to exceed the capabilities of a single CCD. The [[EPYC 7252]] SKU notably has the same specifications as this model except for supporting dual-socket systems and offering 64 MiB L3 cache which evidently requires two CCDs. | ||
== Cache == | == Cache == | ||
Line 47: | Line 53: | ||
|l1d break=8x32 KiB | |l1d break=8x32 KiB | ||
|l1d desc=8-way set associative | |l1d desc=8-way set associative | ||
+ | |l1d policy=write-back | ||
|l2 cache=4 MiB | |l2 cache=4 MiB | ||
|l2 break=8x512 KiB | |l2 break=8x512 KiB | ||
|l2 desc=8-way set associative | |l2 desc=8-way set associative | ||
|l2 policy=write-back | |l2 policy=write-back | ||
− | |l3 cache= | + | |l3 cache=32 MiB |
− | |l3 break= | + | |l3 break=4x8 MiB |
}} | }} | ||
== Memory controller == | == Memory controller == | ||
+ | This model supports up to 8 channels of up to DDR4-3200 memory<ref name="specs">[https://www.amd.com/en/products/cpu/amd-epyc-7232p "AMD EPYC™ 7232P"]. <i>AMD.com</i>. Retrieved October 2020.</ref><ref name="datasheet">[https://www.amd.com/system/files/documents/AMD-EPYC-7002-Series-Datasheet.pdf "AMD EPYC™ 7002 Series Processors: A New Standard for the Modern Datacenter"], AMD Publ. #LE-70002, Rev. 02, April 2020</ref> with a theoretical maximum bandwidth of 25.6 GB/s (≈ 23.84 GiB/s) per channel, but is apparently bandwidth limited by the [[amd/infinity_fabric|IFOP]] links between the I/O die and its only one (two?) compute dies (effective bandwidth ≈ 55 GB/s per CCD at 1.46 GHz FCLK<ref name="isscc2020j-chiplet">Naffziger, Samuel; Lepak, Kevin; Paraschou, Milam; Subramony, Mahesh (2020). <i>2.2 AMD Chiplet Architecture for High-Performance Server and Desktop Products</i>. 2020 IEEE International Solid-State Circuits Conference. pp. 44-45. doi:[https://doi.org/10.1109/ISSCC19947.2020.9063103 10.1109/ISSCC19947.2020.9063103]</ref>). According to AMD this processor is "optimized for 4 channels with DDR4-2667 DIMMs" (≈ 21.3 GB/s per channel) and therefore has a "per-socket theoretical memory bandwidth 85.3 GB/s" (≈ 79.47 GiB/s).<ref name="datasheet"/> | ||
+ | |||
{{memory controller | {{memory controller | ||
|type=DDR4-3200 | |type=DDR4-3200 | ||
Line 71: | Line 80: | ||
== Expansions == | == Expansions == | ||
− | {{expansions | + | {{expansions main |
− | | pcie revision | + | | |
− | | pcie lanes | + | {{expansions entry |
− | | pcie config | + | |type=PCIe |
− | | pcie config 2 | + | |pcie revision=4.0 |
− | + | |pcie lanes=128 | |
− | + | |pcie config=x16 | |
− | + | |pcie config 2=x8 | |
− | + | }} | |
− | |||
− | |||
− | |||
− | |||
− | |||
}} | }} | ||
− | == Features == | + | == Features == |
{{x86 features | {{x86 features | ||
|real=Yes | |real=Yes | ||
Line 106: | Line 110: | ||
|sse42=Yes | |sse42=Yes | ||
|sse4a=Yes | |sse4a=Yes | ||
+ | |sse_gfni=No | ||
|avx=Yes | |avx=Yes | ||
+ | |avx_gfni=No | ||
|avx2=Yes | |avx2=Yes | ||
− | + | |avx512f=No | |
+ | |avx512cd=No | ||
+ | |avx512er=No | ||
+ | |avx512pf=No | ||
+ | |avx512bw=No | ||
+ | |avx512dq=No | ||
+ | |avx512vl=No | ||
+ | |avx512ifma=No | ||
+ | |avx512vbmi=No | ||
+ | |avx5124fmaps=No | ||
+ | |avx512vnni=No | ||
+ | |avx5124vnniw=No | ||
+ | |avx512vpopcntdq=No | ||
+ | |avx512gfni=No | ||
+ | |avx512vaes=No | ||
+ | |avx512vbmi2=No | ||
+ | |avx512bitalg=No | ||
+ | |avx512vpclmulqdq=No | ||
|abm=Yes | |abm=Yes | ||
|tbm=No | |tbm=No | ||
Line 122: | Line 145: | ||
|clmul=Yes | |clmul=Yes | ||
|f16c=Yes | |f16c=Yes | ||
+ | |bfloat16=No | ||
|tbt1=No | |tbt1=No | ||
|tbt2=No | |tbt2=No | ||
|tbmt3=No | |tbmt3=No | ||
+ | |tvb=No | ||
|bpt=No | |bpt=No | ||
|eist=No | |eist=No | ||
Line 130: | Line 155: | ||
|flex=No | |flex=No | ||
|fastmem=No | |fastmem=No | ||
+ | |ivmd=No | ||
+ | |intelnodecontroller=No | ||
+ | |intelnode=No | ||
+ | |kpt=No | ||
+ | |ptt=No | ||
+ | |intelrunsure=No | ||
+ | |mbe=No | ||
|isrt=No | |isrt=No | ||
|sba=No | |sba=No | ||
Line 147: | Line 179: | ||
|securekey=No | |securekey=No | ||
|osguard=No | |osguard=No | ||
+ | |intqat=No | ||
+ | |dlboost=No | ||
|3dnow=No | |3dnow=No | ||
|e3dnow=No | |e3dnow=No | ||
Line 160: | Line 194: | ||
|sensemi=Yes | |sensemi=Yes | ||
|xfr=No | |xfr=No | ||
+ | |xfr2=No | ||
+ | |mxfr=No | ||
+ | |amdpb=No | ||
+ | |amdpb2=Yes | ||
+ | |amdpbod=No | ||
}} | }} | ||
+ | |||
+ | == References == | ||
+ | <references/> |
Latest revision as of 01:29, 19 August 2022
Edit Values | |
EPYC 7232P | |
General Info | |
Designer | AMD |
Manufacturer | TSMC, GlobalFoundries |
Model Number | 7232P |
Part Number | 100-000000081, 100-100000081WOF |
Market | Server |
Introduction | August 7, 2019 (announced) August 7, 2019 (launched) |
Release Price | $450 |
Shop | Amazon |
General Specs | |
Family | EPYC |
Series | 7002 |
Locked | Yes |
Frequency | 3,100 MHz |
Turbo Frequency | 3,200 MHz |
Clock multiplier | 31 |
Microarchitecture | |
ISA | x86-64 (x86) |
Microarchitecture | Zen 2 |
Core Name | Rome |
Core Family | 23 |
Core Model | 49 |
Core Stepping | B0 |
Process | 7 nm, 14 nm |
Technology | CMOS |
MCP | Yes (2 dies) |
Word Size | 64 bit |
Cores | 8 |
Threads | 16 |
Max Memory | 4 TiB |
Multiprocessing | |
Max SMP | 1-Way (Uniprocessor) |
Electrical | |
TDP | 120 W |
Packaging | |
Package | SP3, FCLGA-4094 (FC-OLGA) |
Dimension | 75.4 mm × 58.5 mm × 6.26 mm |
Pitch | 0.87 mm × 1 mm |
Contacts | 4094 |
Socket | SP3, LGA-4094 |
EPYC 7232P is a 64-bit octa-core x86 server microprocessor designed and introduced by AMD in mid-2019. This multi-chip processor, which is based on the Zen 2 microarchitecture, incorporates logic fabricated TSMC 7 nm process and I/O fabricated on GlobalFoundries 14 nm process. The 7232P has a TDP of 120 W with a base frequency of 3.1 GHz and a boost frequency of up to 3.2 GHz. This processor supports single-socket configurations only and up to 4 TiB of memory per socket.
This processor has 8 CPU cores and 32 MiB L3 cache. A single Core Complex Die (plus I/O die) can provide these resources, however there seems to be no evidence that AMD produces any Zen 2-based EPYC or Threadripper processors with an odd number of CCDs, and the advertised maximum memory bandwidth appears to exceed the capabilities of a single CCD. The EPYC 7252 SKU notably has the same specifications as this model except for supporting dual-socket systems and offering 64 MiB L3 cache which evidently requires two CCDs.
Cache[edit]
- Main article: Zen 2 § Cache
Cache Organization
Cache is a hardware component containing a relatively small and extremely fast memory designed to speed up the performance of a CPU by preparing ahead of time the data it needs to read from a relatively slower medium such as main memory. The organization and amount of cache can have a large impact on the performance, power consumption, die size, and consequently cost of the IC. Cache is specified by its size, number of sets, associativity, block size, sub-block size, and fetch and write-back policies. Note: All units are in kibibytes and mebibytes. |
|||||||||||||||||||||||||||||||||||||
|
Memory controller[edit]
This model supports up to 8 channels of up to DDR4-3200 memory[1][2] with a theoretical maximum bandwidth of 25.6 GB/s (≈ 23.84 GiB/s) per channel, but is apparently bandwidth limited by the IFOP links between the I/O die and its only one (two?) compute dies (effective bandwidth ≈ 55 GB/s per CCD at 1.46 GHz FCLK[3]). According to AMD this processor is "optimized for 4 channels with DDR4-2667 DIMMs" (≈ 21.3 GB/s per channel) and therefore has a "per-socket theoretical memory bandwidth 85.3 GB/s" (≈ 79.47 GiB/s).[2]
Integrated Memory Controller
|
||||||||||||||
|
Expansions[edit]
Expansion Options |
|||||
|
Features[edit]
[Edit/Modify Supported Features]
References[edit]
- ↑ "AMD EPYC™ 7232P". AMD.com. Retrieved October 2020.
- ↑ 2.0 2.1 "AMD EPYC™ 7002 Series Processors: A New Standard for the Modern Datacenter", AMD Publ. #LE-70002, Rev. 02, April 2020
- ↑ Naffziger, Samuel; Lepak, Kevin; Paraschou, Milam; Subramony, Mahesh (2020). 2.2 AMD Chiplet Architecture for High-Performance Server and Desktop Products. 2020 IEEE International Solid-State Circuits Conference. pp. 44-45. doi:10.1109/ISSCC19947.2020.9063103
Has subobject "Has subobject" is a predefined property representing a container construct and is provided by Semantic MediaWiki. | EPYC 7232P - AMD#pcie + |
base frequency | 3,100 MHz (3.1 GHz, 3,100,000 kHz) + |
clock multiplier | 31 + |
core count | 8 + |
core family | 23 + |
core model | 49 + |
core name | Rome + |
core stepping | B0 + |
designer | AMD + |
die count | 2 + |
family | EPYC + |
first announced | August 7, 2019 + |
first launched | August 7, 2019 + |
full page name | amd/epyc/7232p + |
has advanced vector extensions | true + |
has advanced vector extensions 2 | true + |
has amd amd-v technology | true + |
has amd amd-vi technology | true + |
has amd precision boost 2 | true + |
has amd secure encrypted virtualization technology | true + |
has amd secure memory encryption technology | true + |
has amd sensemi technology | true + |
has amd transparent secure memory encryption technology | true + |
has ecc memory support | true + |
has feature | Advanced Vector Extensions +, Advanced Vector Extensions 2 +, Advanced Encryption Standard Instruction Set Extension +, SenseMI Technology + and Precision Boost 2 + |
has locked clock multiplier | true + |
has simultaneous multithreading | true + |
has x86 advanced encryption standard instruction set extension | true + |
instance of | microprocessor + |
is multi-chip package | true + |
isa | x86-64 + |
isa family | x86 + |
l1$ size | 512 KiB (524,288 B, 0.5 MiB) + |
l1d$ description | 8-way set associative + |
l1d$ size | 256 KiB (262,144 B, 0.25 MiB) + |
l1i$ description | 8-way set associative + |
l1i$ size | 256 KiB (262,144 B, 0.25 MiB) + |
l2$ description | 8-way set associative + |
l2$ size | 4 MiB (4,096 KiB, 4,194,304 B, 0.00391 GiB) + |
l3$ size | 32 MiB (32,768 KiB, 33,554,432 B, 0.0313 GiB) + |
ldate | August 7, 2019 + |
manufacturer | TSMC + and GlobalFoundries + |
market segment | Server + |
max cpu count | 1 + |
max memory | 4,194,304 MiB (4,294,967,296 KiB, 4,398,046,511,104 B, 4,096 GiB, 4 TiB) + |
max memory bandwidth | 190.7 GiB/s (195,276.8 MiB/s, 204.763 GB/s, 204,762.566 MB/s, 0.186 TiB/s, 0.205 TB/s) + |
max memory channels | 8 + |
microarchitecture | Zen 2 + |
model number | 7232P + |
name | EPYC 7232P + |
package | SP3 + and FCLGA-4094 + |
part number | 100-000000081 + and 100-100000081WOF + |
process | 7 nm (0.007 μm, 7.0e-6 mm) + and 14 nm (0.014 μm, 1.4e-5 mm) + |
release price | $ 450.00 (€ 405.00, £ 364.50, ¥ 46,498.50) + |
series | 7002 + |
smp max ways | 1 + |
socket | SP3 + and LGA-4094 + |
supported memory type | DDR4-3200 + |
tdp | 120 W (120,000 mW, 0.161 hp, 0.12 kW) + |
technology | CMOS + |
thread count | 16 + |
turbo frequency | 3,200 MHz (3.2 GHz, 3,200,000 kHz) + |
word size | 64 bit (8 octets, 16 nibbles) + |