Dell PowerEdge R7615 servers with Broadcom 100GbE NICs can deliver lower-latency, higher-throughput networking to speed your AI fine-tuning tasks

Dell PowerEdge R7615 servers with Broadcom 100GbE
NICs can deliver lower-latency, higher-throughput
networking to speed your AI fine-tuning tasks
A cluster of Dell™
PowerEdge™
R7615 servers featuring AMD EPYC processors
achieved much stronger performance on multi-GPU, multi-node operations using
Broadcom 100GbE NICs than the same cluster using 10GbE NICs
Organizations across industries, from small businesses to Fortune 500 enterprises, are considering how
they can use generative AI (GenAI) to improve their operations. According to a recent McKinsey report,
the pace of technological innovation in this space has been remarkable. During 2023 and 2024, the size
of the prompts that large language models (LLMs) can process, known as “context windows,” spiked
from 100,000 to 2 million tokens.1
This is roughly the difference between adding one research paper to a
model prompt and adding about 20 novels to it. And the types of content that GenAI can process have
continued to increase.
One way to join the GenAI revolution that many organizations are considering is to start with a public
large language model (LLM) and fine-tune it with your own data to build your own in-house LLM. But what
hardware should you choose for the resource-intensive task of training this model? Training an LLM typically
requires the resources of many GPUs. One effective approach is to use a cluster of server nodes, each with
its own set of GPUs, and spread the work across the distributed GPUs. In this environment, low latency and
high bandwidth between GPUs become important. We explored this approach by testing the performance
of a two-node Dell cluster with two networking configurations: one with Broadcom®
100GbE BCM57508
NetXtreme-E network interface cards (NICs) with remote direct memory access (RDMA) over Ethernet
(RoCE) support, and the other with Broadcom 10GbE BCM57414 NICs. The cluster comprised two Dell
PowerEdge R7615 servers with AMD EPYC™
9374F processors and NVIDIA®
L40 GPUs.
LLM training and inference frameworks deployed on distributed GPUs use low-level algorithms to move
data between GPUs, operate on that data, and share the results with other GPUs. Our testing focused on
three of these fundamental algorithms as implemented in the NVIDIA Collective Communications Library
(NCCL) library. This library, which many AI frameworks use, has the advantage of being able to send data
over RoCE network paths or ordinary Ethernet network paths, and it can perform RDMA transfers between
distributed NVIDIA GPUs.
Up to 6.1x the
bandwidth on multi-
GPU, multi‑node
operations*
Up to 83% less time
to complete multi-
GPU, multi‑node
operations*
Up to 66% lower
latency on multi-
GPU, multi-node
operations*
*cluster of Dell PowerEdge R7615 servers featuring AMD EPYC 9374F processors and
Broadcom 100GbE BCM57508 NetXtreme-E NICs vs. the same cluster with 10GbE NICs.
Dell PowerEdge R7615 servers with Broadcom 100GbE NICS can deliver lower-latency,
higher-throughput networking to speed your AI fine-tuning tasks
December 2024
A Principled Technologies report: Hands-on testing. Real-world results.

For each configuration, we studied three multi-GPU, multi-node AI computations from the NCCL test suite2
at packet sizes ranging from 4 B to 256 MB and measured the time to complete the operation and the
effective bandwidth of the network during the operation. This operational bandwidth is a combination of
the very fast data transfer between GPUs on the same node, and the slower data transfer between GPUs on
different nodes. Across this range of packet sizes and each of the three low-level AI operations, the cluster
with 100GbE networking dramatically outperformed the cluster with 10GbE networking. Compared to the
10GbE networking configuration, the operational latency decreased by 26 percent to 67 percent, and the
operational bandwidth was 3.7 to 6.1 times as high. In addition, the 100GbE cluster achieved these gains
without increasing power usage.
Please note that these test do not send enough data between servers to overwhelm the networking link.
Rather, these tests comprise a sequence of computational steps on each GPU, where a given step may
require data from other GPUs. In such cases, a GPU can only start the next computational step once it has
the data from those other GPUs, even if that data is as small as a single byte. The operational bandwidth
depends on the timely transfer of data between GPUs on different servers. The quality of this data transfer
depends on three factors: the time to transfer small amounts of data from a GPU to the server’s NIC, the
time to transfer this data through the network link to the second server’s NIC, and the time to transfer this
data from this NIC to the second GPU.
The value of an in-house LLM for small and medium businesses
AI technologies are complex, and it would be easy to assume that only the largest organizations can
utilize AI effectively and at scale. But that’s not the case. In a recent survey, eight out of ten businesses
with under $1M in revenue reported that they already rely on AI tools.3
According to the Bipartisan
Policy Center, which surveyed businesses on their use of digital tools, “Significant progress in connecting
small business owners to AI has occurred over the last two years.”4
Just as large enterprises are building
AI implementations for everything from product development to customer service, small and medium
businesses (SMBs) are improving business operations using AI.
The idea of a private LLM, trained on your own organization’s existing data and updated regularly as new
data comes in, is particularly appealing. LLMs trained on your own data allow you to gain all the benefits of
an AI chatbot while keeping your data in house, thus maintaining data privacy. SMBs could both save time
and access new opportunities by building and utilizing such LLMs. Manufacturing organizations might be
able to leverage their LLMs to find defects more quickly. Companies across industries could benefit from
LLMs that can analyze images in ways that target specific business needs.
Building an in-house LLM requires a great deal of planning. One of the first steps in the planning process
is selecting the technology solution. You’ll likely want powerful computing resources and networking, and
sourcing them from a manufacturer with significant AI experience could provide further benefits.
Dell: A proven partner for AI
While we highlight the performance of one specific Dell server in this paper, Dell offers a large range of AI
solutions and services. In the 2024 Principled Technologies report “Meeting the challenges of AI workloads
with the Dell AI portfolio,” we highlight advantages Dell brings for AI. According to that report, the Dell AI
portfolio offers “professional and consultative services that help customers build implementation roadmaps
and prepare their data for AI models….training courses that cover machine learning (ML) concepts and
other educational topics…[and] validated designs for AI to help ensure implementation success.”5
December 2024 | 2

Our approach to testing
Training LLMs with custom data typically requires many GPUs, which companies can deploy in a multi-
node cluster. Modern LLM frameworks such as DeepSpeed, Megatron, and PyTorch perform fundamental
arithmetic and data-transfer operations on an LLM spread across all GPUs. Low network latency and high
bandwidth are necessary for performance because, e.g., the overall computation rate slows if GPUs are
waiting for data.
We performed tests to determine the operational latency and throughput for three multi-node, multi-GPU
tasks common to and necessary for LLM data-parallelism methods and LLM model-parallelism frameworks.
We used tasks from NCCL, which uses RoCE, when present, to speed inter-node GPU communications (see
the box “What are RDMA and RoCE?” to learn more). NCCL optimizes GPU communication to achieve
high bandwidth and low latency over PCIe and NVLink high-speed interconnects within a node and across
nodes.6
In our tests, we used publicly available Broadcom driver modules to achieve this functionality, viz.
GPUDirect, for the PCIe and RoCE interconnects.
To assess the benefits of choosing low-latency, high-speed Broadcom NICs, we tested the cluster’s
performance with two network configurations: one with 100GbE Broadcom BCM57508 NetXtreme-E
NICs with RoCE and one with 10GbE NICs. Table 1 provides an overview of the hardware in our test
configurations. For greater detail, including how we configured the network switch for RoCE, see the
science behind the report.
Table 1: The two cluster configurations we tested. Source: Principled Technologies.
100GbE cluster configuration 10GbE cluster configuration
2 x Dell PowerEdge R7615 servers
3 x NVIDIA L40 GPUs per server
1 x AMD EPYC 9374F processor per server
1 x Dell PowerSwitch Z9100-ON (for both 100Gbps and 10Gbps configurations)
Broadcom BCM57508 10/25/50/100/200G
NetXtreme-E NIC with RoCE
Broadcom BCM57414 10G/25G
NetXtreme-E NIC
Broadcom software to extend RDMA
into the NVIDIA GPUs
Note: The Broadcom BCM57508 NetXtreme-E NIC supports the following speeds: 10GbE, 25GbE, 40GbE, 50GbE, 100GbE, and
200GbE. We used the 100GbE setting. The Broadcom BCM57414 10G/25G NetXtreme-E NIC supports the following speeds: 10GbE
and 25GbE. We used the 10GbE setting.
About the Dell PowerEdge R7615
The Dell PowerEdge R7615, featuring 4th
Generation AMD EPYC processors, is a 2U, single-socket server
that Dell has “designed to be the best investment per dollar for your data center.”7
In another recent PT
study, we found that the PowerEdge R7615 could deliver 44 percent better MySQL database performance
than a legacy server, which supports consolidation and the possibility of OpEx savings.8
Learn more at
https://www.delltechnologies.com/asset/en-us/products/servers/technical-support/poweredge-r7615-
spec-sheet.pdf.
December 2024 | 3

Why test the impact of network speed on training?
Much of the AI activity in the news emphasizes the inference stage of the AI LLM workflow. Before
inference, however, comes LLM training. Publicly available AI models are pre-trained on general sets
of data. If organizations wish to use these pre-trained models, they may skip straight to using them for
inference—at the cost of being unable to leverage their own in-house data while maintaining the privacy
of that data. Alternatively, organizations can train the models on their own corpuses of data. This requires
them to go through an additional phase of training, but at the end of that phase, the model could base its
output on an organization’s specific data.
Low-latency networking hardware, such as the Broadcom 100GbE BCM57508 NetXtreme-E NICs with
which we tested, is especially useful in an AI training setting, as we’ll explore in greater detail later
in this report.
For our testing, we chose three multi-GPU, multi-node NCCL primitive operations for AI that are commonly
used in GenAI frameworks performing LLM training with GPUs. Those operations are:
• all-reduce: Operate on the entire dataset, distribute across all GPUs in the cluster, and store the
single result on each GPU
• reduce-scatter: Divide the data on every GPU into logical chunks, and operate on each chunk across
the cluster to form partial results. Then send one partial result to each GPU and store it there
• send-receive: Send data from one GPU to another on the second server, and return a response
What are RDMA and RoCE?
In a multi-node cluster, where each server has its own GPU(s), the speed at which data can travel from one
GPU to another—inter-GPU communications—plays a vital role in performance.
To understand the findings of our testing, it’s useful to know two terms: RDMA and RoCE. Remote Direct
Memory Access, or RDMA, supports moving data from application memory on one server to that on
another server without any CPU involvement. Broadcom describes RoCE as follows: “RoCE (RDMA over
converged Ethernet) is a complete hardware offload feature supported on Broadcom Ethernet network
adapters, which allows RDMA functionality over an Ethernet network. RoCE helps to reduce CPU workload
as it provides direct memory access for applications bypassing the CPU. As the packet processing and
memory access are done in hardware, RoCE allows for higher throughput, lower latency, and lower CPU
utilization on both the sender and the receiver side, which are critical for Machine Learning (ML/AI),
Storage, and High Performance Compute (HPC) applications.”9
The Broadcom NICs we used in our 100GbE cluster configuration support RoCE. In our testing, this direct
memory transfer allowed data to travel efficiently from a GPU on one node to a GPU on another node
without the CPU becoming involved, a significant factor in the performance gains we observed in this
configuration. While the Broadcom BCM57414 10GbE NICs are advanced enough to support RoCE, we
chose not to use then to see how non-RoCE NICs performed in an AI training environment.
December 2024 | 4

What we found
Once you’ve decided to put your own data to use and create a tailored LLM in house, the next step is
deciding which hardware you’ll use to support your LLM. The servers and networking solution you choose
for LLM training should be able to process data quickly to speed up the training process so you can
ultimately move on to the next phase. Better performance means you can complete training operations on
larger data sets faster, and get to a viable AI implementation sooner. As the test results we present in this
section illustrate, the Dell solution with Broadcom BCM57508 10/25/50/100/200G NetXtreme-E can give
organizations the performance they need for in-house LLM training.
Time to complete tasks
Figures 1 through 3 show our multi-GPU, multi-node performance results on the three AI fine-tuning tasks
for the two networking configurations. We see the same pattern across the results: As the size of the data
increased, the amount of time the configuration with the slower 10GbE networking needed increased at a
much faster rate than the configurations with faster 100GbE networking. At the largest packet size we tested,
the 100GbE networking configuration took approximately one-sixth the time to complete each of the tasks
as the 10GbE configuration did. At this size, time to complete decreased by 82 percent on the all-reduce and
reduce-scatter tasks (see Figures 1 and 2) and by 83 percent on the send-receive task (see Figure 3).
Data size (MB)
Operation
time
(microseconds)
0
100,000
200,000
300,000
400,000
500,000
600,000
0 50 100 150 200 250 300
100GbE 10GbE
All-reduce performance: Time to complete task
Lower is better
Figure 1: Performance
of all-reduce multi-GPU,
multi-node task in terms
of time in microseconds
to complete the task on
datasets of multiple sizes.
Lower is better. Source:
Principled Technologies.
About Broadcom BCM57508 10/25/50/100/200G NetXtreme-E NICs
Broadcom network adapters support RoCE, which provides direct memory access for applications, allowing
them to bypass the processor and thus reduce overall CPU load. Skipping the processor can result in
higher throughput, which can speed up AI training workloads. According to Broadcom, the BCM57508
10/25/50/100/200G NetXtreme-E NIC we used in our testing “builds upon the success of the widely-
deployed NetXtreme E-Series architecture by combining a high-bandwidth Ethernet controller with a
unique set of highly-optimized hardware acceleration engines to enhance network performance and
improve server efficiency.”10
To learn more about these NICs, visit
https://www.broadcom.com/products/ethernet-connectivity/network-adapters/bcm57508-200g-ic.
December 2024 | 5

0 50 100 150 200 250 300
Data size (MB)
Operation
time
(microseconds)
Reduce-scatter performance: Time to complete task
Lower is better
0
100,000
200,000
300,000
400,000
500,000
600,000
100GbE 10GbE
Figure 2: Performance of
reduce-scatter multi-GPU,
multi-node task in terms
of time in microseconds
to complete the task on
datasets of multiple sizes.
Lower is better. Source:
Data size (MB)
Operation
time
(microseconds)
Send-receive performance: Time to complete task
Lower is better
0
100,000
200,000
300,000
400,000
500,000
600,000
0 50 100 150 200 250 300
100GbE 10GbE
Figure 3: Performance of
send-receive multi-GPU, multi-
node task in terms of time in
microseconds to complete the
task on datasets of multiple
sizes. Lower is better. Source:
About 4th
Gen AMD EPYC processors
The servers we tested used AMD EPYC 9374F processors, part of the 4th
Gen AMD EPYC processor
family. According to AMD, this group of processors “feature the performance, scalability, compatibility,
and energy efficiency to support hosting of advanced GPU AI engines.”11
EPYC processors include AMD
InfinityGuard, which AMD describes as “a set of layered, cutting-edge security features that help you
protect sensitive data and avoid the costly downtime cause by security breaches.”12
To learn more about 4th
Gen AMD EPYC processors, visit https://www.amd.com/en/products/processors/server/epyc.html.
December 2024 | 6

Latency for multi-GPU, multi-node AI tasks
We measured the latency for the distributed GPU operations by examining the completion time for
small packet size (4 B for all-reduce and send-receiver and 48 B for reduce-scatter), where the on-GPU
computational time was minimal and the inter-GPU communications dominated.13
The latencies we
measured are in Table 1. As it shows, using the 100GbE NIC improved latency by over 65 percent for the
all-reduce and reduce-scatter tasks, and by 26.7 percent for the send-receive task.
Table 2: Latency of multi-GPU, multi-node operations. Lower latency and higher percentage improvement are better. Source:
Multi-GPU,
multinode operation
Latency (microseconds)
Lower is better
Percentage reduction
Higher is better
100GbE configuration 10GbE configuration
all-reduce (packet size: 4 B) 40 123 67.4%
reduce-scatter (packet size: 4 B) 29 85 65.8%
send-receive (packet size: 48 B) 41 56 26.7%
Bandwidth for multi-GPU, multi-node AI tasks
When completing AI training workflows, the rate at which data travels across your AI solution matters. The
greater the flow, or bandwidth, from a GPU on one node to a GPU on another, the better. Choosing AI
solutions with greater bandwidth reduces possible performance bottlenecks and can leave you room to
scale as your AI needs continue to grow.
On all three tasks, we saw dramatically greater bandwidth for multi-GPU, multi-node operations with the
100GbE network configuration. On all-reduce and reduce-scatter tasks, the bandwidth was 5 times that
of the 10GbE configuration (see Figures 4 and 5). On the send-receive task, the 100GbE configuration
achieved 6 times the bandwidth (see Figure 6). Note that in Figure 5, we see one data point at which
bandwidth exceeds 10Gbps on the 10GbE adapter. We believe that intra-GPU traffic—data moving within a
GPU—caused this.
0
10
20
30
40
50
0 50 100 150 200 250 300
Data size (MB)
Bandwidth
(Gbps)
All-reduce bandwidth
Higher is better 100GbE 10GbE
Figure 4: Bandwidth
achieved for multi-GPU,
multi-node all-reduce task.
Higher is better. Source:
December 2024 | 7

0
10
20
30
40
50
0 50 100 150 200 250 300
Data size (MB)
Bandwidth
(Gbps)
Reduce-scatter bandwidth
Higher is better 100GbE 10GbE Figure 5: Bandwidth achieved
for multi-GPU, multi-node
reduce-scatter task. Higher
is better. Note that the
operational bandwidth at
4MB for the 10G network
actually exceeds 10Gbps.
For this packet size, the
speed and amount of data
transferred between GPUs
on one node contributed
more to the operational
bandwidth than that for inter-
node data transfers. Source:
0
10
20
30
40
50
0 50 100 150 200 250 300
Data size (MB)
Bandwidth
(Gbps)
Send-receive bandwidth
Higher is better 100GbE 10GbE
Figure 6: Bandwidth available
for multi-GPU, multi-
node send-receive task.
Higher is better. Source:
About Dell PowerSwitch Z9100-ON Series switches
The Dell EMC Z9100-ON is a 10/25/40/50/100GbE fixed switch Dell has designed specifically to support
applications in high-performance data center and computing environments. The 1RU switch offers a
choice of the following: 32 ports of 100GbE (QSFP28), 64 ports of 50GbE (QSFP+), 32 ports of 40GbE
(QSFP+), 128 ports of 25GbE (SFP28), or 128+2 ports of 10GbE (using breakout cable). These options
let users to conserve rack space, increase footprint density, and migrate more easily to 100Gbps in the
data center core.
Learn more at https://i.dell.com/sites/doccontent/shared-content/data-sheets/en/Documents/dell-
networking-z9100-spec-sheet.pdf.
December 2024 | 8

Power usage
As AI ripples through global news headlines, the world has been paying close attention to the increased
power and cooling that AI workloads require. According to one Scientific American interview on the topic,
“there’s going to be a growth in AI-related electricity consumption”—although “the latest servers are more
efficient than older ones.”14
Selecting servers with increased power efficiency can help you not only reduce
your organization’s carbon footprint but also save on operational expenditures (OpEx), lowering those hefty
power and cooling bills.
We wished to see whether the higher-performing 100GbE environment required more power during our
tests than the 10GbE one. It did not. As we conducted our multi-GPU, multi-node testing, we measured the
power consumption of both servers. Table 3 reports the change to power usage by the two servers at three
representative packet sizes, spanning four orders of magnitude. Despite the great multi-GPU, multi-node
AI task performance improvements the 100GbE Broadcom card enabled, power usage did not increase
significantly with its use. (Note that we did not specifically drill down into GPU power usage during testing;
instead, we report the server’s power usage, which includes the GPU power usage.)
Table 3: Power usage of the two network configurations we tested at three tasks and on three packet sizes: 8KB, 1MB, and
128MB. Lower is better. Source: Principled Technologies.
Power usage by the servers during each test (Watts, Lower is better)
Packet size All-reduce Reduce-scatter Send-receive
100GbE 10GbE 100GbE 10GbE 100GbE 10GbE
8B 1,393.2 1,396.4 1,392.3 1,410.5 1,381.2 1,380.6
1B 1,387.6 1,389.0 1,390.0 1,388.9 1,392.6 1,388.6
128B 1,405.6 1,392.8 1,405.5 1,393.0 1,416.4 1,392.3
The potential for cost savings
Every IT organization has a budget, and according to an Enterprise Technology Research report, IT budget
growth is beginning to slow.15
When you’re considering purchasing a new technology solution to initiate
or grow your AI implementation, you must consider its cost alongside its value. That cost is more than
purchase price. Expenditures for power, cooling, and licensing are also factors in the total cost of a solution
over its lifetime.
Additionally, choosing a solution that can process your in-house data quickly lets you build and fine-tune
your AI model in less time and puts your data to work faster, thereby increasing your business agility and
allowing you to reap the benefits of your AI implementation sooner.
December 2024 | 9

Conclusion
Many companies want to do LLM training on their internal data so they can use it to solve a host of
business problems. LLM training uses low-level fundamental operations over distributed GPUs. When these
operations perform efficiently, your LLM training takes much less time to complete and you can have your
AI implementation operational sooner. Our tests looked at the performance of fundamental operations
over distributed GPUs. We found that using Broadcom 100GbE BCM57508 NICs in a cluster of two Dell
PowerEdge R7615 servers with AMD EPYC processors and NVIDIA GPUs provided dramatically lower
latency and greater bandwidth than using only 10GbE networking, with no increase in power usage.
1. “McKinsey Technology Trends Outlook 2024,” ac-
cessed November 19, 2024, https://www.mckinsey.
com/capabilities/mckinsey-digital/our-insights/the-
top-trends-in-tech#new-and-notable.
2. NVIDIA, “NCCL Tests,” accessed November 16,
2024, https://github.com/NVIDIA/nccl-tests.
3. Small Business & Entrepreneurship Council, “Small
Business AI Adoption Survey October 2023,” ac-
cessed November 17, 2024, https://sbecouncil.org/
wp-content/uploads/2023/10/SBE-Small-Business-AI-
Survey-Oct-2023-FINAL.pdf.
4. Sujan Garapati, “Poll Shows Small Businesses Are
Interested in and Benefit from AI,” accessed No-
vember 17, 2024, https://bipartisanpolicy.org/blog/
poll-shows-small-businesses-are-interested-in-and-
benefit-from-ai/.
5. “Meeting the challenges of the Dell AI portfolio,” ac-
cessed November 17, 2024, https://www.principled-
technologies.com/Dell/AI-portfolio-vs-HPE-0124.pdf.
6. NVIDIA, “NVIDIA Collective Communications Library
(NCCL),” accessed November 21, 2024, https://de-
veloper.nvidia.com/nccl.
7. “PowerEdge R7615,” accessed November 17, 2024,
https://www.delltechnologies.com/asset/en-us/prod-
ucts/servers/technical-support/poweredge-r7615-
spec-sheet.pdf.
8. “Improve performance and gain room to grow by
easily migrating to a modern OpenShift environment
on Dell PowerEdge R7615 servers with 4th
Genera-
tion AMD EPYC processors and high-speed 100GbE
Broadcom NICs,” accessed November 17, 2024,
https://www.principledtechnologies.com/clients/
reports/Dell/PowerEdge-R7615-100GbE-Broadcom-
NICs-MYSQL-database-0524/index.php.
9. Broadcom, “RDMA over Converged Ethernet
(RoCE),” accessed November 21, 2024,
https://techdocs.broadcom.com/us/en/stor-
age-and-ethernet-connectivity/ethernet-nic-con-
trollers/bcm957xxx/adapters/RDMA-over-Con-
verged-Ethernet.html.
10. Broadcom, “BCM57508 - 200GbE,” accessed
November 21, 2024, https://www.broadcom.com/
products/ethernet-connectivity/network-adapters/
bcm57508-200g-ic.
11. AMD, “Nothing Stacks up to EPYC,” accessed
November 25, 2024, https://www.amd.com/en/prod-
ucts/processors/server/epyc.html.
12. AMD, “Nothing Stacks up to EPYC.”
13. NVIDIA, “Performance reported by NCCL tests,”
accessed November 15, 2024, https://github.com/
NVIDIA/nccl-tests/blob/master/doc/PERFORMANCE.
md.
14. Lauren Leffer, “The AI Boom Could Use a Shocking
Amount of Electricity,” accessed November 17, 2024,
https://www.scientificamerican.com/article/the-ai-
boom-could-use-a-shocking-amount-of-electricity/.
15. ETR, “2024 IT Budget Growth is Slowing,” ac-
cessed November 17, 2024, https://www.linkedin.
com/pulse/2024-budget-growth-slowing-etr-enter-
prise-technology-research-t9ikf/.
Principled Technologies is a registered trademark of Principled Technologies, Inc.
All other product names are the trademarks of their respective owners.
For additional information, review the science behind this report.
Principled
Technologies®
Facts matter.®
Principled
Technologies®
Facts matter.®
This project was commissioned by Dell Technologies.
Read the science behind this report at https://facts.pt/cbYq3Wb
December 2024 | 10

Dell PowerEdge R7615 servers with Broadcom 100GbE NICs can deliver lower-latency, higher-throughput networking to speed your AI fine-tuning tasks

More Related Content

Similar to Dell PowerEdge R7615 servers with Broadcom 100GbE NICs can deliver lower-latency, higher-throughput networking to speed your AI fine-tuning tasks (20)

More from Principled Technologies (20)

Recently uploaded (20)

Dell PowerEdge R7615 servers with Broadcom 100GbE NICs can deliver lower-latency, higher-throughput networking to speed your AI fine-tuning tasks